<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Ruwanthi (Ru) Somatilaka on Medium]]></title>
        <description><![CDATA[Stories by Ruwanthi (Ru) Somatilaka on Medium]]></description>
        <link>https://medium.com/@ruwanthisulanjali?source=rss-6ccfc057c5e0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*QW6YISgzmRiemF9JHrx3yA@2x.jpeg</url>
            <title>Stories by Ruwanthi (Ru) Somatilaka on Medium</title>
            <link>https://medium.com/@ruwanthisulanjali?source=rss-6ccfc057c5e0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 09:35:52 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@ruwanthisulanjali/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[60 Usability Issues, One Turning Point, and What I Should Have Known Earlier]]></title>
            <link>https://ruwanthisulanjali.medium.com/60-usability-issues-one-turning-point-and-what-i-should-have-known-earlier-46579e8b2e4d?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/46579e8b2e4d</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[web-design]]></category>
            <category><![CDATA[usability-testing]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Fri, 27 Mar 2026 05:56:49 GMT</pubDate>
            <atom:updated>2026-03-27T06:20:44.543Z</atom:updated>
            <content:encoded><![CDATA[<p>A real case study on AI as a teammate, priority frameworks, and the cost of reviewing mid-development. <em>Project by </em><a href="https://www.linkedin.com/company/dziel-creative-ltd/"><em>DZiel Creative LTD</em></a><em>.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OD3f8hoFAYJpWEx-Icdz2g.png" /></figure><h3>The Context</h3><p>This is my third project with the same client. He found me in the middle of his first project, and we’ve worked together since. Each time, the dynamic is the same, a separate team handles the design and development, and I come in as the UX lens from the outside.</p><p>The client has a multi-sport indoor facility in Melbourne offering tennis, futsal, basketball, netball, pickleball, cricket and soccer. The site covers court bookings, league competitions, party bookings, and general facility information, seven pages in total with a lot of moving parts.</p><p>The client’s requirement was straightforward, <strong>find the issues in their design</strong>. But what I actually conducted goes beyond issue-spotting. A <strong>usability evaluation</strong> assesses how easy and enjoyable it is for users to achieve their goals while using a product.</p><p>That means looking at the design not just for visual or technical problems, but through the lens of the people who will actually use it, someone trying to book a court, a parent registering their child for a junior league, a team captain checking the competition schedule, or a party organiser looking for a venue.</p><p>Every finding in the audit is rooted in one question: <strong><em>can the user achieve what they came here to do, and how friction-free is that experience?</em></strong></p><figure><img alt="UX design usability evaluation by DZiel Creative Ltd showing prioritisation framework, issues analysis, and lessons learned from failed approach" src="https://cdn-images-1.medium.com/max/1024/1*psonnB1fDKFVnMBZaEDw1w.png" /><figcaption>Tried Differently, Failed, and Learned</figcaption></figure><h3>What Was Done Differently This Time</h3><p><a href="https://medium.com/@ruwanthisulanjali/case-study-a-client-website-was-reviewed-before-launch-heres-what-was-found-d98d7bc53f1b">On previous projects</a> <em>(February 2026)</em>, I was reviewing development sites before they went live, meaning the build was already done and feedback at that stage was expensive to act on. This time I asked for the <strong>Figma file upfront</strong>, before anything went to the developer.</p><p>The idea was simple, catching issues at the design stage is dramatically cheaper and faster than catching them after development. A misplaced CTA in Figma is a five-minute fix. The same issue found after a developer has built the page, written the logic, connected the links, fixed user flows could take hours to unpick.</p><blockquote><em>[ What Went Wrong ]</em></blockquote><blockquote><strong>But here’s the honest reality though.</strong> Even though I had the Figma file, the development team had already started building in parallel to meet a deadline. So while asking for the file earlier was the right instinct, the timing still meant some feedback landed mid-build rather than before it.</blockquote><h4>[ What Was Learned ]</h4><p>The lesson I’m taking forward is this, asking for the file is only half the battle. The other half is <strong>agreeing upfront on a review checkpoint before development starts</strong>. Without that agreement, you can have the Figma file and still be too late.</p><p>Let’s walk through the steps taken to complete the project, the challenges faced, and the different approaches used along the way.</p><figure><img alt="AI-assisted UX design workflow showing four-phase process where AI acts as a team member in usability analysis, prioritisation, and design decision-making" src="https://cdn-images-1.medium.com/max/1024/1*PZ-TuJNBsvW5dNwY4jdfXg.png" /><figcaption>Working with AI</figcaption></figure><h3>Step 1: My Own Analysis First</h3><p>In a previous article <em>(July 2025)</em>, I explored <strong>why I don’t lead with AI when doing UX audits</strong>.</p><blockquote><a href="https://medium.com/design-bootcamp/should-ux-designers-trust-ai-with-user-behaviour-analysis-53c9ec629f9c">Should UX Designers Trust AI with User Behaviour Analysis?</a></blockquote><p>The short version is that AI-generated audit documents can lack consistency with the actual visuals, miss important context, and produce generic observations that don’t reflect what’s really on the screen. I want indepth analysis which AI fails to do at the moments.</p><p><strong>So my process always starts with my own analysis</strong>. I went through every page (homepage, sport pages, about us, parties, contact, junior competitions, adult competitions) and documented what I found independently before involving any AI tool.</p><h3>Step 2: AI as a Teammate</h3><p>After completing my own analysis, I gave the AI the same screenshots and asked for its perspective. This is where it got interesting.</p><h4>Where AI aligned with me</h4><p>Most of the core usability issues I had flagged were confirmed. Also noted the broken booking CTA, the weak “Read More” labels, the footer missing key navigation items.</p><p>Having a second opinion that aligned gave me more confidence in flagging these to the client.</p><h4>Where AI caught things I missed</h4><p>There were a few observations I hadn’t noted, particularly around <strong>content consistency</strong> across pages.</p><p>One thing expert reviews are known for is blind spots, you develop a pattern of looking for certain things and can miss others. AI doesn’t have that same pattern bias, which made it genuinely useful as a second pass.</p><h4>Where AI was surface-level</h4><p>Some of the points AI raised were correct in principle but <strong>lacked the depth needed to be actionable</strong>. For example, AI flagged “low contrast text” in forms but didn’t connect that to the specific legal and financial risk those conditions of entry carried.</p><p>That contextual reasoning (understanding <em>why</em> something matters beyond the visual) <strong>still required human judgement</strong>.</p><blockquote>The way I think about it: AI is a thorough first reviewer who catches a lot but doesn’t always understand the stakes. My job is to add that layer of stakes and context.</blockquote><h3>Step 3: Didn’t Work as Planned, So Prioritisation Was Required</h3><p>In my <strong>first version</strong> of the audit, I delivered the findings as issues and recommendations. Clear, structured, and detailed.</p><blockquote>[ What Went Wrong ]</blockquote><blockquote>But then the development team reviewed the audit, and because they hadn’t considered usability in the original design stage and they already started the developments, the findings felt like a significant amount of additional work.</blockquote><blockquote>They hadn’t budgeted for it and quote requested to increase. Which is extra cost for my client.</blockquote><p>This is a dynamic I’ve seen before and it’s worth naming honestly, when <strong>usability is not part of the design conversation from the start</strong>, it lands as scope creep when raised later. Even when the issues are legitimate and important, the timing makes them feel like additions rather than corrections.</p><h4>[ How the Issue Was Resolved ]</h4><p>To resolve this, I went back to the audit document and added two things.</p><ul><li>First, I numbered every observation (60 items across 7 pages). The document was structured to be read column by column rather than in a Z-reading pattern, so the numbers follow that same flow, making it easier to scan annotated pages without losing your place.</li><li>Second, the observations were organised into key categories: <strong>Usability, Content Updates, Visual Design Updates, </strong>and<strong> Missing Elements</strong>. This helped structure the findings and made them easier to navigate for both the client and the development team.</li></ul><p>With the findings now structured and categorised, the next step was to prioritise which issues needed to be addressed first, and that’s where AI came in again.</p><figure><img alt="Home page usability review with UX design evaluation, content structure analysis, and accessibility considerations" src="https://cdn-images-1.medium.com/max/1024/1*u6XViCznSkWWVA2IEPQlDA.png" /><figcaption>Home page analysis</figcaption></figure><h4>[ What Was Learned ]</h4><p>Based on the categorisation, a prioritisation framework was then planned to guide implementation, ensuring that the most impactful issues could be addressed first.</p><p>This changed the conversation. Instead of a list of problems with no sense of urgency or order, the team now had a clear working document, fix these before go-live, address these in the first two weeks, schedule the rest. The scope felt manageable because it was staged.</p><h3>Step 4: Prioritising with AI, and Where It Needed to Be Challenged</h3><p>With 60 numbered items across 7 pages, prioritising everything manually would have taken significant time. So I gave the numbered list to AI and asked it to sort items into four tiers:</p><ul><li><strong>Critical (</strong>blocks a user action or conversion)</li><li><strong>High Priority (</strong>damages trust or causes confusion leading to drop-off)</li><li><strong>Medium Priority</strong> (affects quality but doesn’t block the journey)</li><li><strong>Low Priority (</strong>cosmetic or post-launch enhancement)</li></ul><blockquote>This saved me a lot of time. But I didn’t just accept the output.</blockquote><figure><img alt="UX priority list of usability issues categorised by severity levels including critical, high, medium, and low priority design improvements" src="https://cdn-images-1.medium.com/max/1024/1*e8rBwMO7DQO1JJIiOmWgAA.png" /><figcaption>Priority list of usability issues.</figcaption></figure><h4>Where AI Was Challenged</h4><p>There were several places where I disagreed with the AI’s categorisation and changed the priority level. For example, the logo colour change on scroll was initially placed as Low Priority, which I challenged immediately. While I wasn’t certain it deserved Critical status, it was clearly not Low Priority either.</p><p>Two reasons AI completely missed:</p><ul><li>First, the logo appears on every single page, so an invisible white logo against a white nav background on scroll looks broken site-wide, not just on one page.</li><li>Second, and more importantly, the logo is the client’s brand identity. A logo that disappears or clashes on scroll isn’t just a visual glitch, it undermines brand credibility across the entire site. I moved it to Medium Priority.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/952/1*-ZpYpgrH0cMdNexV75LISg.png" /><figcaption>Challenging AI Through Conversation</figcaption></figure><p>The prioritisation process became a conversation rather than a handover. I’d review AI’s reasoning, <strong>apply my own judgement about business impact and context, and adjust where the logic didn’t hold</strong>.</p><p>That back and forth actually sharpened my own thinking about why each issue sat where it did.</p><h3>What I Found, Key Issues Across the Site</h3><p>Here’s a summary of the most significant findings, drawn from the full 60-item audit.</p><h4>#1 Critical: No Court Booking Form Exists and Confusing User Flow.</h4><figure><img alt="UX usability issue in court booking system showing confusing external link affecting user navigation and booking flow experience" src="https://cdn-images-1.medium.com/max/1024/1*AI18Pr9kFfUYzXW9TwCQlg.png" /><figcaption>Court Booking: Confusing External Link</figcaption></figure><p>The primary CTA “Book a Court” appears on the homepage, every sport page, and the navigation, but there is no internal court booking form anywhere on the site. <strong>Every page promises a booking experience the site cannot currently deliver</strong>.</p><p>The old site had an external booking link for tennis only, and the system was labelled “Indoor Tennis”, which is already confusing for a multi-sport facility offering futsal, basketball, netball, pickleball, and soccer. That solution never worked for all sports, and carrying it forward into the new site would create the same problem at a larger scale.</p><p>This is the single most important issue to resolve before launch. Two options were recommended:</p><ul><li><strong>Option 1: </strong>Build an internal booking form within the site, keeping users in the same brand experience from first click to confirmation.</li><li><strong>Option 2: </strong>Recreate a dedicated booking system that covers all sports, which is the client’s preferred direction.</li></ul><p>Whichever approach is taken, the “Book a Court” CTA must lead somewhere that delivers on its promise, otherwise every page on the site is setting an expectation the product cannot meet.</p><h4>#2 Critical: Registration Process Has No Payment Step and Does Not Complete the User Flow</h4><figure><img alt="Critical UX issue in registration process showing missing payment step causing incomplete user flow and failed conversion journey" src="https://cdn-images-1.medium.com/max/1024/1*BoTjR5SMKGWsBfR3Nb1S5w.png" /><figcaption>Form Critical Issues</figcaption></figure><blockquote>Note for Image: This form had many usability and content issues identified in the full audit. Only the most significant findings are highlighted here.</blockquote><p>The team registration form collects team details across three steps but has no payment step at the end. Users complete the form with no idea when or how to pay the $90 registration fee or even whether their registration has been confirmed.</p><p><strong>Three critical gaps</strong> make this process incomplete:</p><ul><li><strong>Payment: </strong>there is no payment step after “Register My Team.” Users submit the form without any indication of how or when the $90 fee will be collected, creating confusion and potential no-shows.</li><li><strong>Player registration form: </strong>the conditions of entry reference a separate player registration form, but there is no link, download, or explanation of where to find it. A parent reading this has no idea what to do next.</li><li><strong>Individual player journey: </strong>the form states it should be completed by the team captain only, but there is no guidance on how an individual player without a team can find or join one.</li></ul><p>Most critically, there is no confirmation of what happens after clicking “Register My Team.” No confirmation message, no next steps, no follow-up email mentioned. The user journey simply stops, which means potential no-shows and unnecessary admin burden for the venue.</p><h4>#3 Critical: Conditions of Entry Fail Accessibility Standards</h4><figure><img alt="Critical UX accessibility issue in conditions of entry page failing accessibility standards and creating barriers for inclusive user experience" src="https://cdn-images-1.medium.com/max/1024/1*QquCemxsJp0Vp4BinwE_HQ.png" /><figcaption>Critical form issues</figcaption></figure><p>The conditions of entry on the registration form include financial commitments ($90 per game, $90 registration, $90 forfeit fee) displayed in grey text on a dark background that fails basic <strong>accessibility contrast standards</strong>. Users are legally agreeing to financial terms they may not be able to read.</p><p>Three CSS fixes are needed before launch: increase text to full white, increase font size, and add spacing between each condition once the font size is increased.</p><h4>#4 High Priority: Wrong CTAs on the About Us Page</h4><figure><img alt="High priority UX issue on About Us page showing incorrect call-to-action (CTA) placement affecting user navigation and conversion flow" src="https://cdn-images-1.medium.com/max/1024/1*iZDvVM6maGreKAbeaAcWIQ.png" /><figcaption>High-Priority Issues Identified on the About Us Page</figcaption></figure><blockquote>Note for Image: This page had many usability, visual design, content, and missing elements identified in the full audit. Only the most significant finding is highlighted here.</blockquote><p>The About Us hero has “Book My Party” and “Party Plans” as its primary and secondary CTAs. Someone landing on About Us is in trust-building mode (they want to learn who you are before committing). Pushing a party booking at that exact moment is jarring and misplaced. These CTAs belong on the Parties page, not here.</p><p>This appears to be a content placement oversight, the party CTAs likely carried over from another page without considering the intent of someone landing on About Us.</p><p>The <strong>recommendation</strong> is to replace them with CTAs that direct users back into the conversion flow once trust has been established, something like “Book a Court” or “Explore Our Sports.”</p><h4>#5 High Priority: No Timetable on Adult Competition Pages</h4><figure><img alt="High priority UX issue showing missing timetable on adult competition pages affecting user experience and event information accessibility" src="https://cdn-images-1.medium.com/max/1024/1*WonL8O4zsUggOQyOCLLBLQ.png" /><figcaption>High-Priority Issues Identified on the Competition Pages</figcaption></figure><p>The junior competitions page handles this well, it has a detailed day-by-day schedule showing age groups and time slots, and the “Time Table” CTA in the hero section directs users straight to it. The user flow is clear.</p><p>The three adult competition pages, however, have the same “Time Table” CTA in the hero but nowhere to direct users, because no timetable exists on those pages. The CTA promises something the page cannot deliver.</p><p>A team captain needs to know when games are played before they commit to registering. <strong>Without that information on the page, the registration form becomes a much harder sell</strong>.</p><h3>The Priority Framework: A Lesson I Should Have Applied Earlier</h3><p><strong><em>One of the biggest friction points</em></strong> in this project came when the developer reviewed the audit and felt the scope of changes was overwhelming. The developer came back asking to increase the quote, which meant extra cost for the client.</p><p>Looking back, the issue wasn’t the number of findings. It was that everything initially carried the same visual weight. A low-priority copyright text fix looked as significant as a critical broken user journey. No wonder it felt overwhelming.</p><blockquote>I introduced the four-tier priority framework to resolve this. And it worked.</blockquote><p>Once the developer could see that only a handful of items needed to be fixed before launch, and everything else was staged across the weeks and months after, the conversation changed.</p><figure><img alt="High priority UX issue showing missing timetable on adult competition pages affecting user experience, clarity, and event information accessibility" src="https://cdn-images-1.medium.com/max/1024/1*VT-1_Y3MhdLFVlJ__-dBTw.png" /><figcaption>Four Tier Priority Framework Proposed by DZiel Creative</figcaption></figure><p>The framework I’d now use on every audit:</p><blockquote><strong><em>Critical: </em></strong><em>no launch without this fixed</em></blockquote><blockquote><strong><em>High Priority: </em></strong><em>fix within 1 weeks of launch</em></blockquote><blockquote><strong><em>Medium Priority: </em></strong><em>fix within the first month</em></blockquote><blockquote><strong><em>Low Priority: </em></strong><em>backlog, fix when capacity allows</em></blockquote><h4><strong>[ The honest reflection ]</strong></h4><p>I should have built this framework into the audit from day one, not added it in response to pushback. Delivering 60 findings without a clear staging plan puts all the anxiety on the client and developer.</p><p><strong>The priority framework is what makes a long audit feel manageable rather than overwhelming</strong>.</p><h3>Working Across Separate Teams</h3><p>One dynamic worth naming, on this project, the design team and development team are different people, and I sit outside both. That creates a communication challenge that the audit document has to navigate carefully.</p><p>Design decisions were made without full development context. Development pushed back on design without always having the UX rationale. And feedback from an <strong>outside UX reviewer can land as a third voice adding noise rather than clarity</strong>.</p><p>What I found helped was writing the audit document to speak to both audiences simultaneously, framing visual issues visually, framing development issues in terms of effort and risk, and letting the priority framework do the bridging work. Both teams need to know what matters most and in what order, regardless of who owns the fix.</p><h3>What I’d Do Differently Next Time</h3><ul><li><strong>Agree on a review checkpoint before development starts</strong>, not just ask for the Figma file. The file without a freeze on development doesn’t give you much leverage.</li><li><strong>Deliver the priority framework on day one</strong>, not as a response to pushback. It sets healthier expectations from the start and protects the relationship with the development team.</li><li><strong>Use AI earlier in the prioritisation process</strong>, but always review and challenge the output. It saves significant time on large audits but needs human judgement applied on top.</li></ul><h3>Final Thought</h3><p>The most valuable thing I delivered on this project wasn’t the list of 60 issues. <strong>It was the framework that made those 60 issues actionable</strong>.</p><p><strong><em>Anyone can find problems</em></strong>. The skill is helping a team understand which problems to solve first, why, and what the consequences are if they don’t. That’s what turns a UX audit from a document people read once into something that actually shapes how a product gets built.</p><h3>Interested in a Usability Evaluation for Your Website?</h3><p>Whether you’re preparing to launch or looking to improve an existing site, a professional usability evaluation can uncover issues that are costing you conversions, and give you a clear, prioritised roadmap to fix them.</p><blockquote><strong>Get in touch with </strong><a href="https://www.linkedin.com/company/dziel-creative-ltd/"><strong>DZiel Creative</strong></a><strong> to find out more.</strong></blockquote><p><em>DZiel Creative Ltd supports businesses through UX evaluations, website design, UI/UX, branding, consulting, and mentoring. The focus is on understanding users, simplifying complexity, and designing digital experiences that feel intuitive and purposeful. Follow along on </em><a href="https://ruwanthisulanjali.medium.com/"><em>Medium</em></a><em> for more case studies and insights.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=46579e8b2e4d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[[Case Study] A Client Website Was Reviewed Before Launch. Here’s What Was Found.]]></title>
            <link>https://ruwanthisulanjali.medium.com/case-study-a-client-website-was-reviewed-before-launch-heres-what-was-found-d98d7bc53f1b?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/d98d7bc53f1b</guid>
            <category><![CDATA[web-design]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[conversion-optimization]]></category>
            <category><![CDATA[case-study]]></category>
            <category><![CDATA[usability]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Mon, 23 Feb 2026 07:26:44 GMT</pubDate>
            <atom:updated>2026-02-23T07:35:11.138Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A full </em><strong><em>usability evaluation</em></strong><em> of an Australian sport centre website, 40+ issues, clear recommendations, and a client relationship that’s still going. Project by DZiel Creative LTD.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*01vS6hm6BA796M_rtlOBeA.png" /></figure><h3>The Project Brief</h3><p>Not every project starts with a design brief. Some start with a conversation, a website that’s almost ready to go live, and a client who wants to make sure they’re putting their best foot forward before the world sees it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/988/1*nebuvU94z1WfzKhQXLTCjA.png" /><figcaption>First client message received by DZiel Creative regarding a new project enquiry</figcaption></figure><p>That’s exactly how this one began.</p><p>A client based in Australia, running an indoor sports and trampoline centre, was days away from launching a new website. Before going live, they wanted a professional set of eyes on it.</p><p>Not to redesign it. Not to rebuild it from scratch. <strong>Just to catch what they might have missed, and make sure users would actually be able to find what they were looking for and convert</strong>.</p><p><a href="https://www.linkedin.com/company/dziel-creative-ltd/">DZiel Creative</a> was brought in to conduct a full usability evaluation across the entire site.</p><h3>What Is a Usability Evaluation, Exactly?</h3><p>Before getting into the findings, it’s worth explaining what a usability evaluation actually is, because clients often aren’t sure.</p><p>A usability evaluation assesses how easy and enjoyable it is for users to achieve their goals while using a product. It can be carried out at any stage of design or development, and there are multiple methods available depending on the project goals, timeline, and resources (<em>Interaction Design Foundation</em>, 2016).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*euVZx4QlFu7nF6nG0lHAtA.png" /><figcaption>Overview of key findings and recommendations from a usability evaluation conducted by DZiel Creative.</figcaption></figure><p>For this project, a usability inspection method was used, specifically an <strong>expert review</strong>. This involves an experienced UX professional assessing the site to identify usability problems, evaluating it against <strong>usability goals</strong>, <strong>accessibility standards</strong>, <strong>design principles</strong>, and <strong>heuristics</strong>. Unlike usability testing, it doesn’t involve live users, which makes it particularly well suited for pre-launch reviews where the goal is to catch issues before real users ever encounter them (<em>Interaction Design Foundation</em>, 2016).</p><p>In this case, the evaluation went beyond pure usability. <strong>Foundational SEO and CRO (conversion rate optimisation)</strong> knowledge was applied throughout, and coding knowledge was used to inspect the underlying code, identifying technical errors and flagging areas where development best practices hadn’t been followed. This gave the client a much fuller picture than a visual review alone would have provided.</p><p>The goal isn’t to criticise.</p><blockquote>The goal is to identify anything that could confuse users, prevent conversions, or create a poor experience, and then provide clear, actionable recommendations to fix it.</blockquote><p>For this client, the evaluation covered five key pages: the Home page, the Party page, the Contact Us page, the About Us page, and the Gallery page. In total, over 40 individual issues were identified and documented across the site.</p><h3>A Note on AI and This Evaluation</h3><p>This evaluation was conducted entirely through human expert review, no AI tools were used to analyse user behaviour.</p><p>There were a few reasons for this.</p><figure><img alt="Human-led usability evaluation conducted without AI tools" src="https://cdn-images-1.medium.com/max/1024/1*Ly5jMo25UAvI4OGaKKr37A.png" /><figcaption>Usability evaluation was completed entirely through expert human review, without using AI tools to analyse user behaviour.</figcaption></figure><p>First, the site wasn’t live yet. The development link wasn’t publicly accessible, which means most AI scanning tools wouldn’t have been able to reach it in the first place.</p><p>Second, when AI has been used in past evaluations, feeding screenshots and asking for an analysis, the results needed to be double-checked for inconsistencies. <strong>That back-and-forth actually took more time than doing the analysis directly</strong>. When the knowledge and experience is already there, adding an extra layer of verification in the middle doesn’t make the process more efficient.</p><p>Third, this evaluation required going beyond surface-level observations. Testing forms, checking interactions, tracing user flows step by step, all with a focus on lead generation. <strong>That level of depth isn’t something current AI tools can reliably replicate</strong>. Doing it manually was simply faster and more accurate.</p><blockquote>The goal has always been to deliver documentation that goes deep, focusing on CRO and real usability impact, not just a list of observations that looks good on paper.</blockquote><blockquote>If you’re curious about the wider conversation around AI and UX analysis, this topic was explored in an earlier article: <a href="http://ruwanthisulanjali.medium.com/should-ux-designers-trust-ai-with-user-behaviour-analysis-53c9ec629f9c">Should UX Designers Trust AI with User Behaviour Analysis?</a></blockquote><h3>The Challenge: Writing for Two Very Different Audiences</h3><p>One of the biggest challenges of this project had nothing to do with the website itself.</p><p><strong>It was the evaluation document.</strong></p><figure><img alt="Visual explaining clear notes for both client understanding and developer action" src="https://cdn-images-1.medium.com/max/1024/1*JMM2wHvq91cLB3GkA3QfNg.png" /><figcaption>Feedback should clearly explain the issue and its impact for clients, while also giving precise instructions for developers on what to fix.</figcaption></figure><p>The client had <strong>no UX background</strong>. The development team, on the other hand, <strong>needed technical clarity</strong>. Writing one document that could genuinely serve both audiences, without dumbing things down for the developers or overwhelming the client with jargon, required real intentionality.</p><p>Every finding was written with plain language first, followed by a clear rationale. Not “this violates Nielsen’s heuristic of consistency and standards”, but “this is confusing because users expect the navigation and the footer to match, and right now they don’t.”</p><blockquote>The goal was simple: the client should be able to read it on their own and understand exactly what was wrong and why it mattered. The developers should be able to pick it up and know exactly what to fix.</blockquote><h3>The Findings: What Was Uncovered</h3><p>In total, over 40 issues were identified across five pages, ranging from broken mobile interactions and missing legal requirements to confusing user flows and SEO gaps that would have hurt the site before it even had a chance to perform.</p><p>Rather than listing every single finding here, the focus is on the ones that tell the most interesting story.</p><h4>1. The icon that looked like a gun</h4><p>The “Why Choose Us” section featured an infographic, a great idea for communicating value quickly. But the icon chosen to represent “Supervised Fun” looked, unmistakably, like a gun.</p><p>It wasn’t intentional. These things happen when designs are reviewed in isolation rather than through fresh eyes. But on a family-focused indoor sports website aimed at children’s birthday parties, it was exactly the kind of detail that needed catching before launch.</p><p>The section also had a busy background image placed behind the infographic text, making the content almost impossible to read. “Why Choose Us” is a high-value section for conversions, but the design was working against it.</p><figure><img alt="Screenshot of “Why Choose Us” section showing an icon resembling a gun and unreadable text" src="https://cdn-images-1.medium.com/max/1024/1*UkiS2TasYTIFpH95LuFjgA.png" /><figcaption>Infographic in the “Why Choose Us” section uses an icon that looks like a gun and has text hard to read due to a busy background</figcaption></figure><h4>2. The party page that couldn’t book a party</h4><p>On the Trampoline Party page, the main CTA directed users to “Book a Jump Pass.” On a page specifically about booking a party, that’s the wrong destination entirely. Users who had already decided they wanted a party were being sent somewhere else, directly disrupting the lead generation user flow.</p><p>It didn’t stop there. The other two party pages on the site had a party booking form built right in. The Trampoline Party page had none. Users who made it to that page had no direct way to book, an obvious and easily fixable conversion gap.</p><figure><img alt="Screenshot of Trampoline Party page showing missing party booking form and incorrect CTA" src="https://cdn-images-1.medium.com/max/1024/1*keViipFflZ-ZEZlRI2RLMw.png" /><figcaption>Trampoline Party page lacks a booking form and has a CTA directing users away, disrupting conversions</figcaption></figure><h4>3. Red used as a warning, on a party page</h4><p>Color carries meaning, and red in particular is psychologically associated with danger, warnings, and stop signals. On the party page, red was used to highlight a note reminding users to bring their own cake.</p><p>The intention was helpful but the experience felt like a warning.</p><p>This was part of a wider color issue across the site. Red made up approximately 60% of the visual weight, against white at 30% and blue at 10%. Good visual design follows the 60–30–10 rule (<em>Interaction Design Foundation</em>, 2017), but the dominant colour should support the experience. Here, it was working against it, particularly on mobile, where the effect was even more intense.</p><figure><img alt="Party page and Home page showing red used for warnings, dominating visual weight over other colors" src="https://cdn-images-1.medium.com/max/1024/1*9xLn8WXtLX8I3-eEZiBuaQ.png" /><figcaption>Party and home page showing red used for cake reminder, appearing as a warning.</figcaption></figure><h4>4. The CTA that disappeared on mobile</h4><p>The homepage had a clear “Contact Us” CTA on desktop. On mobile, it was gone entirely.</p><p>This matters because the majority of users today browse on their phones. A missing CTA on mobile is a missing conversion opportunity. Several other elements were also broken or misaligned on mobile, suggesting the mobile experience hadn’t been thoroughly reviewed before the site was prepared for launch.</p><figure><img alt="Missing “Contact Us” CTA and broken elements on mobile homepage from DZiel Creative analysis" src="https://cdn-images-1.medium.com/max/1024/1*-befeFhMiss_WQe_KzHO0w.png" /><figcaption>Mobile homepage with missing “Contact Us” CTA and broken elements</figcaption></figure><h4>5. SEO issues that would have hurt from day one</h4><p><strong>This wasn’t a design only evaluation</strong>. Foundational SEO knowledge was applied throughout, and what was found was significant.</p><p>Not a single page on the site had an H1 tag. Every page needs exactly one, it’s how search engines understand what a page is about. Without it, the site was essentially invisible to Google from a structural standpoint. Multiple images were also missing ALT tags, affecting both SEO and accessibility. And a data collection form was live without a privacy policy notice, a legal requirement when collecting personal information like email addresses and phone numbers.</p><p>These weren’t edge cases. They were issues that would have had real consequences from the moment the site went live.</p><figure><img alt="website SEO issues: missing H1, missing image ALT tags, and no privacy notice on form." src="https://cdn-images-1.medium.com/max/1024/1*PTFM4Ms9XdLHwCDYnufY4Q.png" /><figcaption>Highlights key SEO problems: missing H1s, missing ALT text, and privacy policy absent on a data form.</figcaption></figure><h4>6. The small things that add up</h4><p>Beyond the headline issues, there were dozens of smaller findings, hover effects on non-clickable elements that confused users, external links opening in the same tab causing the site to disappear, navigation showing two party options in one place and three in another, a contact form that allowed submission without a message, and a gallery carousel on mobile that worked once and then stopped responding entirely.</p><p>Individually, each of these feels minor. <strong>Together, they create an experience that feels unpolished and erodes user trust</strong>, even if users can’t explain exactly why.</p><blockquote>These are just the highlights. The full evaluation report covers every finding in detail, with annotated screenshots and specific recommendations for each page.</blockquote><blockquote>If you’d like to see what a full report looks like, feel free to reach out to me and one will be sent across.</blockquote><h3>The First Meeting</h3><p>After completing the evaluation, DZiel Creative presented the findings to the client. The document was structured to walk through each page clearly, with every issue explained in plain language, supported by visual callouts and screenshots, and paired with a specific recommendation.</p><p>The client’s response was immediate and genuine:</p><blockquote>“Just read the whole PDF audit you have done, love the attention to detail!! I will definitely have to get you on a call with the guys and lets go through this together. There’s one more website coming too to audit, will send that after we finalise this one. Thanks for all the work.” — Mr. V. Company Director.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/988/1*fZtENGjTwEZRJjGU3wddBQ.png" /><figcaption>Customer review for DZile Creative Ltd</figcaption></figure><p>That message said a lot. Not just that the client was happy, but that they trusted the process enough to want to bring the development team in, and that they were already thinking about the next project.</p><h3>What Happened Next</h3><p>The development team reviewed the evaluation and began implementing the recommended changes. A follow-up review was conducted once the updates were made, and while many issues had been addressed, there were still significant items outstanding that needed to be resolved before the site could responsibly go live.</p><p>This is normal. A usability evaluation isn’t a one-and-done exercise. Real improvement is iterative.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Gt8K8gOZbwtPVu4RrdVFFg.png" /><figcaption>Visual illustrating the iterative process of usability evaluation, implementation of recommendations, follow-up review, and ongoing refinements before final launch.</figcaption></figure><p>The site is still in the pre-launch phase, with final fixes being worked through before the go-live date. DZiel Creative remains involved, and a meeting with the development team is scheduled to walk through the remaining items together.</p><p>And true to the client’s word, a second website project has already been confirmed.</p><h3>Beyond the Evaluation: Planning for Real User Data</h3><figure><img alt="DZiel Creative proposal outlining heatmap setup for future data analysis" src="https://cdn-images-1.medium.com/max/1024/1*06D6BZwvPUFtOtnP5WOR3g.png" /><figcaption>DZiel Creative recommending the implementation of heatmaps to track user behaviour and support future data analysis.</figcaption></figure><p>One of the <strong>recommendations</strong> made to the client went beyond the evaluation itself.</p><p>Once the site goes live, the real learning begins. Even the most thorough evaluation is based on expert judgement, it can’t replace data from real users actually using the site. That’s why the client was advised to set up heatmaps and behaviour tracking tools post-launch</p><p>These tools capture things like where users click, how far they scroll, where they drop off, and which CTAs get the most attention. Over time, that data paints a picture of what’s actually working and what still needs improvement, and that’s where the next phase of UX work begins.</p><blockquote>This approach has already delivered results in previous work. In an earlier project, heatmap analysis and real user data were used to increase a law firm’s homepage conversions by 400%. You can read the full case study here: <a href="https://ruwanthisulanjali.medium.com/how-ux-design-increased-a-law-firms-homepage-conversions-by-400-2f05b48649ad"><em>How UX Design Increased a Law Firm’s Homepage Conversions by 400%</em></a></blockquote><p>The goal isn’t just to fix a website before launch. It’s to build something that keeps getting better.</p><h3>What This Project Taught Me</h3><p>A few things stood out from this engagement that are worth reflecting on.</p><h4><strong>1. Identifying problems isn’t enough.</strong></h4><p>Every issue in the evaluation was paired with a recommendation. Telling a client something is wrong without telling them what to do about it isn’t helpful, it’s just stressful.</p><blockquote>UX professionals have a responsibility to close that loop.</blockquote><h4><strong>2. Communication is part of the work.</strong></h4><p>Writing an evaluation document that works for a non-technical client and a development team simultaneously is genuinely difficult. Getting that balance right matters as much as the quality of the findings themselves.</p><h4><strong>3. Trust is built through detail.</strong></h4><p>The client didn’t respond positively because the document was long. They responded because it was <strong>thorough, clear, and showed that real thought had gone into every page</strong>. That attention to detail is what turned a one-off project into an ongoing relationship.</p><h4><strong>4. The job doesn’t end at launch.</strong></h4><p>A website going live is the beginning of the UX journey, not the end. Real user data is the next chapter.</p><h3>Interested in a Usability Evaluation for Your Website?</h3><p>Whether you’re preparing to launch or looking to improve an existing site, a professional usability evaluation can uncover issues that are costing you conversions, and give you a clear, prioritised roadmap to fix them.</p><p><strong>Get in touch with </strong><a href="https://www.linkedin.com/company/dziel-creative-ltd/"><strong>DZiel Creative</strong></a><strong> to find out more.</strong></p><p>DZiel Creative Ltd supports businesses through UX evaluations, website design, UI/UX, branding, consulting, and mentoring. The focus is on understanding users, simplifying complexity, and designing digital experiences that feel intuitive and purposeful. Follow along on <a href="https://ruwanthisulanjali.medium.com/">Medium</a> for more case studies and insights.</p><h3>Reference</h3><p><em>IxDF — Interaction Design Foundation</em>. (2016, June 6). What is Usability Evaluation?. <em>IxDF — Interaction Design Foundation</em>. Available at: <a href="https://www.interaction-design.org/literature/topics/usability-evaluation">https://www.interaction-design.org/literature/topics/usability-evaluation</a> (Accessed: 22 February 2026)</p><p><em>IxDF — Interaction Design Foundation</em>. (2017, July 1). What is Color in UX Design?. <em>IxDF — Interaction Design Foundation</em>. Available at: <a href="https://www.interaction-design.org/literature/topics/color">https://www.interaction-design.org/literature/topics/color</a> (Accessed: 22 February 2026)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d98d7bc53f1b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Human or AI? Measuring the impact of explanation style in an educational support system]]></title>
            <link>https://medium.com/design-bootcamp/human-or-ai-measuring-the-impact-of-explanation-style-in-an-educational-support-system-aefb6e985362?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/aefb6e985362</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[usability]]></category>
            <category><![CDATA[emotional-intelligence]]></category>
            <category><![CDATA[human-ai-collaboration]]></category>
            <category><![CDATA[ux-research]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Fri, 15 Aug 2025 18:52:51 GMT</pubDate>
            <atom:updated>2025-08-16T16:36:09.939Z</atom:updated>
            <content:encoded><![CDATA[<h4>A pilot Wizard-of-Oz study on usability, emotional response, and user perception in Blackboard Ultra</h4><p><em>— This research was conducted in 2023 and updated in 2025 —</em></p><figure><img alt="A person who touches the digital screen with a green code background" src="https://cdn-images-1.medium.com/max/1024/1*g8hpbDv45JRT3CYkaH6qmw.jpeg" /><figcaption>Photo by <a href="https://www.pexels.com/photo/man-between-screens-5198392/">Mohammad Yasir</a></figcaption></figure><h3><strong>Abstract</strong></h3><p><strong>Artificial Intelligence (AI)</strong> is a rapidly expanding field, evidenced by the proliferation of new AI tools. Public perceptions of AI, however, remain diverse, with some users expressing trust and enthusiasm while others remain skeptical. For designers, researching AI can be challenging, as many lack the technical expertise required to develop AI systems.</p><blockquote>This study proposes evaluation methods that enable designers to investigate AI systems while also capturing public perceptions.</blockquote><p>The <strong>research focuses</strong> on Blackboard Ultra, the learning platform used at Glasgow Caledonian University (GCU), which presents usability challenges for new users. An AI-powered chatbot is proposed as a potential solution; however, before implementation, it is essential to determine whether human or AI-generated explanations are more effective in supporting problem-solving.</p><p>The study examines the effects of human–human and human–AI explanations on task performance and perceptions of AI.</p><blockquote><em>Initial findings suggest that </em><strong><em>AI explanations can reduce task completion time but may introduce confusion</em></strong><em>, while </em><strong><em>human explanations often offer emotional comfort but are not always technically efficient</em></strong><em>.</em></blockquote><blockquote><em>A </em><strong><em>Wizard-of-Oz (WoZ)</em></strong><em> approach was employed to simulate AI behaviour, combining pre- and post-test interviews, questionnaires, and observational methods. Participant emotions were analysed using video recordings mapped to the circumplex model of affect.</em></blockquote><p>Due to ethical constraints, a pilot study was conducted with three participants, and the findings were used to inform further research design.</p><h3>1. Introduction and background</h3><p>Artificial intelligence underpins a broad range of applications, from large-scale data analysis to decision-making and conversational systems. Tools such as ChatGPT, built on the GPT-4 architecture, have demonstrated potential to enhance education through personalised, interactive support.</p><p>This study explores the use of AI-driven explanations to improve the user experience (UX) of new students navigating Blackboard Ultra at Glasgow Caledonian University (GCU).</p><p>As building a fully functional AI system was beyond the project’s scope, a Wizard-of-Oz (WoZ) approach was used to simulate an AI capable of detecting user emotions and providing problem-solving guidance, drawing on GPT-4 outputs and Blackboard Help Centre content.</p><p>A <strong>pilot study</strong> with <strong>three participants</strong>, including an <strong>expert evaluator</strong>, revealed inefficiencies in current human–human problem-solving processes and highlighted diverse perceptions of AI. While some participants expressed distrust, others viewed AI as a valuable tool for support. These findings informed the methodology for the subsequent stages of the research.</p><h3>2. Literature review</h3><p>Usability testing has become a widely adopted method in both industry and academia for identifying user experience (UX) issues when developing interfaces and systems (Fan, Shi, and Truong, 2020). Fan <em>et al</em>. (2022) conducted research by simulating an AI with reasonable performance using a <strong>Wizard-of-Oz (WoZ) approach</strong>. They carried out a mixed-methods study involving <strong>24 UX evaluators</strong>, who identified UX issues from usability test videos with the assistance of an AI tool.</p><p>The <strong>results</strong>, both qualitative and quantitative, indicated that AI with explanations, whether presented asynchronously or synchronously, provided better support for UX evaluators’ analyses and was perceived more positively. Furthermore, synchronous AI improved evaluators’ performance and engagement more than asynchronous AI. While Fan <em>et al</em>. (2022) developed the AI specifically for their study, the current research relies entirely on the WoZ approach due to technical limitations.</p><figure><img alt="Screenshot of the AI Assistant interface showing a video player, annotation panel, detected problem table, AI-suggested problem timeline, and AI explanations, 2022" src="https://cdn-images-1.medium.com/max/931/0*ZCymDiBRNpusBmjC" /><figcaption><strong>Figure 1</strong>: Screenshot (<em>The user interface of AI Assistant that presents the AI-suggested problems synchronously with explanations</em>, 2022) of the AI Assistant user interface (a) video player; (b) annotation panel; © detected problem table; (d) AI-suggested problem timeline; and (e) AI-suggested problem explanations.</figcaption></figure><figure><img alt="Research data showing satisfaction, understanding, and trust" src="https://cdn-images-1.medium.com/max/1024/1*V_NjnOhHN04hpI0vCVlcpQ.png" /></figure><p>Vodrahalli <em>et al</em>. (2022) examined how users respond to AI advice by recruiting 1,100 crowd workers. Their experiments led to the development of the activation–integration model for advice utilisation.</p><figure><img alt="Diagram showing participant’s initial responses, based on human advice, AI advice, and participant’s final response" src="https://cdn-images-1.medium.com/max/1024/0*feCwmT1lsXQpsTyP" /><figcaption><strong>Figure 3</strong>: Screenshot (Visualization<em> of our study design</em>, 2022) of the study plan</figcaption></figure><p>Vodrahalli <em>et al</em>. (2022) study revealed that participants’ perceptions of human versus AI performance influenced their decisions to follow recommendations. Interestingly, participants tended to use advice similarly, regardless of whether it came from a human or AI source. The two-stage “activation–integration” model proposed by Vodrahalli <em>et al</em>. (2022) explains human behaviour in the context of advice utilisation and identifies factors shaping human–AI interactions. However, the study did not provide clear conclusions about how participants’ responses evolved after receiving AI explanations.</p><figure><img alt="Data table showing covariates, activation coefficients, and response change coefficients" src="https://cdn-images-1.medium.com/max/1024/1*XkDQEWxwOJNaRfBxEkwHhw.png" /></figure><p>Collectively, these studies highlight the role of AI in supporting evaluation processes and shaping human trust in AI recommendations. However, neither study explicitly examines the direct impacts on individuals following AI-assisted problem-solving, particularly in technical support scenarios.</p><p><strong>This gap</strong> underscores the need to investigate how AI explanations influence user performance, perception, and emotional responses in practical contexts.</p><h3>3. Research question and hypothesis</h3><p>The <strong>aim of this study</strong> is twofold: to propose an <strong>evaluation method for designers</strong> who lack the technical skills to develop AI, and to <strong>gain insights into public perceptions of AI</strong>. Additionally, the study explores the potential of an AI system capable of perceiving <strong>human emotions and providing support</strong> to students using Blackboard. Specifically, the study addresses the following research questions (RQs):</p><p><strong><em>RQ 1</em></strong>: What impact or effect does the AI explanation have on solving the problem in Blackboard?</p><p><strong><em>RQ 2</em></strong>: What impact or effect does the human explanation have on solving the problems in Blackboard?</p><p><strong><em>RQ 3</em></strong>: How does the explanation of the human-human and human-AI situations impact or affect the user’s impression of AI?</p><h4><strong>3.1 Hypothesis</strong>:</h4><ul><li>Human explanations require more time to solve problems and can lead to user frustration.</li><li>AI explanations are faster and can result in higher user satisfaction.</li><li>Users relying on AI explanations will solve problems more quickly than those relying on human explanations.</li><li>Users with a positive initial perception of AI are likely to maintain it, whereas users with negative preconceptions may develop a more favourable view after experiencing AI-assisted support.</li></ul><h3>4. Project methods and design</h3><h4>4.1 Participants</h4><p>Three participants were recruited from the ATHCI module at Glasgow Caledonian University for this pilot study. One served as the <em>human–human explainer</em>, assisting the other two participants in problem-solving tasks. The session lasted <strong>30 minutes</strong> and included both <strong>pre- and post-test interviews</strong>.</p><blockquote>Ethical approval was granted by the Glasgow Caledonian University Human Ethics Committee, and participants provided signed consent via email to the module leader.</blockquote><p>Each participant completed two tasks (Appendix<strong> </strong>A) to evaluate problem-solving effectiveness when using either AI or human assistance. Participants initially indicated their preferred assistance type and could seek help from only one source at a time.</p><p>To <strong>counterbalance preferences</strong>, Participant 2 stated their choice, while the moderator assigned them the opposite option to Participant 1’s selection.</p><h4>4.2 Measures</h4><p>This study employed both<strong> quantitative</strong> and <strong>qualitative</strong> measures to assess the effectiveness of AI and human assistance in problem-solving tasks.</p><ul><li><strong>Performance time:</strong> The duration taken to complete each task, whether with human or AI explanations, was recorded, as well as the time taken before a participant finished the task. The moderator tracked timings, and video recordings were later reviewed to evaluate task completion.</li><li><strong>User satisfaction and frustration:</strong> These were assessed through pre- and post-test interviews and questionnaires, as changes in these measures are closely linked to shifts in user perception.</li><li><strong>Verbalisation of thought process:</strong> While the <em>think-aloud</em> protocol is common in usability testing (Lazar, Feng, &amp; Hochheiser, 2009), it was not adopted here due to prior findings that slow verbalisation can extend task completion times and influence results. Instead, participants were encouraged to share their thoughts periodically, without moderator prompts, to avoid interference with performance measurements.</li></ul><h4>4.3 Method map</h4><figure><img alt="Method map illustration including consent form, pre-test and post-test questionnaires, user observation, testing, and neasures" src="https://cdn-images-1.medium.com/max/612/1*vVzpmz119x1jka26PcjYow.png" /><figcaption><strong>Figure 5</strong>: Method map for the study</figcaption></figure><h4>4.4 Procedures: [Phase 1] Built the test’s foundation</h4><figure><img alt="Phase 1 procedure diagram showing the three steps" src="https://cdn-images-1.medium.com/max/622/0*XHBYAYNTp-I0hlbs" /><figcaption><strong>Figure 6</strong>: Plan of Step 1</figcaption></figure><h4><strong>a. Identify the problem and document it.</strong></h4><p>The <strong>first phase</strong> of this study focused on identifying key issues related to first-time use of Blackboard and establishing the foundation for the tasks. Previous studies (Fan <em>et al</em>., 2022; Fan, Shi, &amp; Truong, 2020) employed qualitative and quantitative methods, such as interviews, questionnaires, and observations , to gather data. However, ethical constraints limited the applicability of these approaches.</p><p>To address this, the study began by documenting past experiences and incidents encountered by <strong>first-time Blackboard users</strong> (Appendix B), forming the basis of Task 1.</p><p><strong>Participants familiar with Blackboard</strong> then created a new task (Task 2) with support from the module leader: setting up a group in the ATHCI module tab and posting a blog, an activity previously unattempted by class members.</p><p>Also <strong>experimented with ChatGPT prompts</strong> and the <strong>Blackboard Help Center</strong> to explore problem-solving strategies.</p><h4><strong>b. </strong>Experiment with ChatGPT, and the Blackboard Help Center.</h4><p>Recent research (Fan et al., 2022) highlights the use of AI measured via the Wizard-of-Oz (WoZ) approach. This method was applied in the current study, following expert recommendations (Ramaswamy, 2023; Porcheron, Fischer, &amp; Reeves, 2021), especially when developing full AI technology is costly or the problem space is broad.</p><p>To maintain realism of the simulation, pre-compiled replies were prepared, reducing delays before the “wizard” responded (Lazar, Feng, &amp; Hochheiser, 2009, p. 276). GPT-4 and the Blackboard Help Center were used as sources of answers to address key issues, as summarised in Table 1.</p><figure><img alt="Table showing Task 1 problems and chat gap responses, and Task 2 with Blackboard Help Center content" src="https://cdn-images-1.medium.com/max/1024/1*asGKrB2B8c_wbxLgyknBpQ.png" /><figcaption><strong>Table 1</strong>: Document the task and find an AI solution. Recent issues encountered by new Blackboard users and have GPT-3 respond to solve those issues. [<a href="https://docs.google.com/spreadsheets/d/1omzbT0RGldzl-imIB_CpAPyruhNTDnk0dmoVpQEY9Lk/edit?gid=0#gid=0">Link</a>].</figcaption></figure><h4>4.5 Procedures: <strong>[Phase 2] Pilot plan</strong></h4><p>To accurately capture <strong>user emotions</strong> during the test, a <strong>mini-emotion lab was established</strong>. The study required session and screen recording as well as a chat platform, and therefore, MS Teams was selected.</p><p>The author of the study, who also served as the moderator and AI, explained the tasks to participants while observing their behaviour through the video call. Following Ramaswamy (2023), the person <strong>acting as AI should remain hidden from participants until the end of the test</strong>, situated outside the room (Figure 9). Although the name of the author appeared on the chat, all participants were familiar with the process and aware that the moderator was also acting as AI.</p><figure><img alt="Phase 2 pilot test plan illustration showing the classroom environment and the outside-the-room layout" src="https://cdn-images-1.medium.com/max/1024/1*VeJZC4yxb0zdmWco2YFl1Q.png" /></figure><h4><strong>a. Preparation</strong></h4><p>Prior to the test day, participants were contacted with reminders to ensure readiness (Lazar, Feng, &amp; Hochheiser, 2009, p. 271). The designated spaces were prepared with printed materials and refreshments (Figures 10, 11, and 12). A moderator protocol was created (Appendix C) to ensure consistency and completeness of the interviews (Kuniavsky, 2003), including reminders to record participants’ responses.</p><figure><img alt="Three images showing the test environment setup for two participants and the moderator" src="https://cdn-images-1.medium.com/max/1024/1*N6YGZ4G9zNlvhMmwVwF6Gw.png" /></figure><p>Payment and refreshment procedures were clarified in advance (Lazar, Feng, &amp; Hochheiser, 2009). Refreshments were offered after the test to prevent pre-test excitement from biasing results. Only the caller received pre-test refreshments due to a longer participation time.</p><h4><strong>b. Participant 1 and the caller</strong></h4><p>Originally planned in the C110 lab, the session was relocated to C117 due to noise and crowding, with internal access to C116A (Figure 9).</p><p>First, a participant who acted as a <strong>caller was invited to the C117 lab</strong> (Figure 12), given a brief introduction about the test, and asked to complete pre-test questionnaires (Appendix D) to identify his current mood and familiarity with Blackboard. Then <strong>participant 1 was invited</strong> <strong>to the C116A room</strong> (Figure 11), where they did the same process and asked for a pre-test questionnaire (Appendix E) to learn about the participant’s current mood, their familiarity with Blackboard, their previous experience with user testing, and their most comfortable option to get help from a human or AI.</p><p>To begin this process, participant 1 was assisted in setting up their computer, after which the moderator left the room. The participant connected via MS Teams and shared their screen (Figure 13). Full screen sharing was required to capture the AI interface; failure to do so previously led to missing data during the experiment sessions.</p><figure><img alt="Participant 1 joining the call, sharing their screen, and displaying the Blackboard interface in the test day" src="https://cdn-images-1.medium.com/max/1024/0*DaHcO34lIVXnsfET" /><figcaption><strong>Figure 13</strong>: Screenshot of the prototype — Participant 1’s screen share after opening Blackboard</figcaption></figure><p><strong><em>Task 1: </em></strong>The moderator explained Task 1 (Appendix A), and the participant 1 had <strong>one minute to attempt the task</strong>. The moderator/AI observed emotional responses. After one minute, AI assistance was offered, and timing was recorded via session analysis (Figure 14).</p><blockquote>AI responses were provided by ChatGPT-recorded answers.</blockquote><p>Once participants completed the task with AI-to-human guidance, the session was marked as completed (“Y”). A semi-structured interview (Appendix F) followed to capture participant reflections. Task 1 results were later compared with Task 2.</p><figure><img alt="Flow diagrams showing the tasks 1, the steps followed by Participant 1, and how task completion time was measured" src="https://cdn-images-1.medium.com/max/1024/0*ZCAV_ZSyfQOjQUiL" /><figcaption><strong>Figure 14</strong>: Participant 1, Task 1, Initial Plan</figcaption></figure><p><strong><em>Task 2: </em></strong>Participant 1 then completed Task 2, which was finished within one minute. The moderator instructed the participant to seek <strong>human assistance from the caller</strong>. The human-human interaction was recorded, and timing was recorded (“X”).</p><p>Post-task interviews captured participants’ experiences, overall test impressions, and preferred support method. Refreshments were provided after completion, and the participant exited the premises (Figures 15).</p><figure><img alt="Flow diagrams showing the tasks 2, the steps followed by Participant 1, and how task completion time was measured" src="https://cdn-images-1.medium.com/max/1024/1*c67hR3G4mltEnaaiQ-SiHw.png" /><figcaption><strong>Figure 15</strong>: Participant 1, Task 2, Initial Plan</figcaption></figure><h4><strong>c. Participant 2 and the caller</strong></h4><p>After Participant 1 departed, <strong>Participant 2 was invited</strong> to ensure no overlap. The same procedure was followed, except Participant 2 received AI assistance from the Blackboard Help Center rather than ChatGPT.</p><blockquote>Task order and support type were <strong>counterbalanced</strong> based on pre-test questionnaire preferences.</blockquote><figure><img alt="Table showing participants, first and second task titles, and the counterbalancing order" src="https://cdn-images-1.medium.com/max/1024/1*BbnuAy-nOn5wcK-mP1J06A.png" /><figcaption><strong>Table 2</strong>: Plan the study counterbalance</figcaption></figure><figure><img alt="Participant 2 joining the call, sharing their screen, and displaying the Blackboard interface in the test day" src="https://cdn-images-1.medium.com/max/1024/0*QpaaqCDSEVKSVES_" /><figcaption><strong>Figure 16</strong>: Screenshot of the Prototype — Participant 2 screen shared after opening Blackboard</figcaption></figure><p>After completing tasks and interviews, refreshments were offered, and Participant 2 exited the premises.</p><h4>4.6 Data gathering and analysis</h4><p>Data for this study were collected through a combination of <strong>pre-test and post-test interviews</strong>, <strong>continuous moderator observations</strong>, and <strong>video analysis</strong> of the session recordings. Participants’ emotions were documented throughout the testing process.</p><blockquote>These observations were then mapped onto Russell’s (1980) <strong>circumplex model</strong> of affect, which considers two dimensions: arousal and valence.</blockquote><p>To enhance the reliability of the emotional analysis, the mapped data were subsequently reviewed and compared by an expert evaluator. In this study, the caller served as the expert evaluator, conducting the review a few days after the test sessions. This process ensured that the interpretation of participants’ emotional responses was both systematic and validated.</p><h3>5. Project execution and results</h3><p>This section presents the data gathered from the pilot project. Both <strong>qualitative and quantitative</strong> approaches were employed to support the study’s hypothesis. Results are summarised at the beginning of each subsection, highlighting how human performance and emotions were influenced by AI assistance during task completion.</p><h4>5.1 Caller pre-test analysis</h4><p>The caller’s pre-test responses were analysed using qualitative and quantitative data. Initial questions assessed the caller’s mood, familiarity with Blackboard, and comfort in providing technical support. As shown in Table 3, the <strong>caller was an expert user familiar with the Blackboard interface</strong>. Although not feeling well, he indicated that this would not affect his ability to assist participants. His previous experience made <strong>him comfortable providing technical support, positively contributing</strong> to the study.</p><p>Emotions were categorised based on <strong>voice tone, moderator observation, and participant feedback</strong>. Overall, the caller exhibited mixed but predominantly pleasant emotions prior to the test.</p><figure><img alt="Table showing caller pre-test questions, recorded emotions, and participant answers as data" src="https://cdn-images-1.medium.com/max/1024/1*XJ0M-gBsr_sE9xYE3Kl-yw.png" /><figcaption><strong>Table 3</strong>: Caller pre-test questionnaire</figcaption></figure><h4>5.2 Participants’ pre-test analysis</h4><p>Participants’ pre-test results (Table 4) highlighted differences in mood, Blackboard experience, and initial perceptions of AI. Both <strong>participants reported good moods and daily Blackboard usage</strong>, suggesting positive task efficiency. Both were intermediate users, indicating similar expertise levels.</p><p>Questions assessing prior perceptions of AI revealed that both participants had neutral impressions (rated at 3 out of 5), with comfort in receiving AI assistance. During testing, participants followed planned counterbalance choices. Emotions were tracked via voice tone, moderator observation, and participant responses.</p><blockquote>Participant 1 had a positive perception of AI, whereas Participant 2 was less satisfied, likely due to prior experience.</blockquote><figure><img alt="Table showing pre-test questions for Participants 1 and 2, recorded their emotions, and participants’ answers as data" src="https://cdn-images-1.medium.com/max/1024/0*ity-ww5H6wrRw5gj" /><figcaption><strong>Table 4</strong>: Pre-test questions, answers from both participants, and how the author feels the emotions in each stage</figcaption></figure><h4>5.3 Participant 1 test results</h4><p>Figure 17 summarises the timeline for Participant 1’s task completion. Notably, it took 1 minute 12 seconds to notice the AI message and give permission, compared with 22 seconds to contact the caller.</p><blockquote>Human explanations took longer to solve tasks (5 minutes 17 seconds) than AI explanations (3 minutes 24 seconds). Despite familiarity with the task, Participant 1 struggled with Task 2 but completed it quickly with AI support.</blockquote><figure><img alt="Timeline illustrating Participant 1’s task completion, including observed behaviors and time taken for each step" src="https://cdn-images-1.medium.com/max/1024/0*OPmHwcl5IMns5QUq" /><figcaption>Figure 17: Timeline of the first participant’s task completion</figcaption></figure><figure><img alt="Diagram showing Participant 1’s task completion times for AI-assisted and human-assisted tasks" src="https://cdn-images-1.medium.com/max/1024/1*JfFiYhR2WLLHA5l3_Cf3Aw.png" /></figure><h4>a. Observation: Participant 1</h4><p>During the test, the moderator (AI) monitored Participant 1 at each stage to identify points of difficulty and activate AI assistance as needed. After the test, session recordings were analysed, and emotional responses at each stage were documented, as shown in Figures 19 and 20.</p><p>Participant 1 received AI support first, and closer examination of their emotional data (Figure 19) indicated that they were mostly neutral throughout the process. The moderator, acting as a Wizard of Oz (WoZ), observed the participant’s behaviour, language, mouse movements, and clicks to detect signs of confusion, then sent the first AI message accordingly.</p><figure><img alt="Diagram showing Participant 1’s emotional changes during AI-human explanations while completing the task 1" src="https://cdn-images-1.medium.com/max/1024/1*dtHZKNpO9DaYZh8UvRLndA.png" /><figcaption><strong>Figure 19</strong>: Participant 1, Emotional Changes in AI-Human Explanation, and Solving the Task</figcaption></figure><p>When assigned Task 2, which was unfamiliar, Participant 1 experienced some confusion (Figure 20). However, they were able to complete the task within the allocated time.</p><p>Emotions expressed during the conversation with the human caller were not recorded, as they did not reflect genuine responses; Participant 1 already possessed the knowledge to solve the task. Consequently, these emotions were excluded from the study’s scope.</p><figure><img alt="Diagram showing Participant 1’s emotional changes during AI-human explanations while completing the task 2" src="https://cdn-images-1.medium.com/max/1024/1*Zfmq6nUTiimoXEtiv81FcQ.png" /><figcaption><strong>Figure 20</strong>: Participant 1, Emotional Changes in Human-Human Explanation, and Solving the Task</figcaption></figure><h4>5.4 Participant 2 test results</h4><p>Figure 21 provides an overview of the summary statistics for the time taken to complete each task.</p><blockquote>A notable observation from Figure 22 is that AI-human explanations required a longer duration (4 minutes and 3 seconds) to complete a task, <strong>whereas human-human explanations were completed more quickly</strong> (1 minute and 15 seconds).</blockquote><p>During Task 1, Participant 2 was expected to finish within 1 minute. However, they chose to decline AI assistance and worked independently, successfully completing the task in 2 minutes and 12 seconds (Figure 22). This duration was not included in the task completion times, as the analysis only considered data from human-human explanations.</p><figure><img alt="Timeline illustrating Participant 2’s task completion, including observed behaviors and time taken for each step" src="https://cdn-images-1.medium.com/max/1024/0*vRfsbq0ucl_p7ojs" /><figcaption><strong>Figure 21</strong>: Timeline of the second participant’s task completion</figcaption></figure><figure><img alt="Diagram showing Participant 2’s task completion times for AI-assisted and human-assisted tasks" src="https://cdn-images-1.medium.com/max/1024/1*EXu0UHxde5m6EQ1GHnP3ng.png" /></figure><h4>a. Observation: Participant 2</h4><p>During the test, the moderator (AI) employed the same observation techniques as described previously. Participant 2’s behaviour throughout the task completion process was notable, as she exhibited a considerable range of emotions (Figures 23 and 24). To ensure accuracy, the moderator, acting as a wizard, assessed her behaviour, mouse movements, and clicks before providing any AI assistance.</p><p>Significant emotional changes were also observed following the participant’s completion of a prior task and interaction with the caller. However, emotions were only recorded during the independent completion of Task 1, as the moderator had obtained a clearer understanding of the participant’s emotional responses before proceeding to Task 2, which involved activating AI assistance in response to user confusion.</p><figure><img alt="Diagram showing Participant 2’s emotional changes during AI-human explanations while completing the task 1" src="https://cdn-images-1.medium.com/max/1024/1*B-5Z0vMr4T5We5YbKY1C7A.png" /><figcaption><strong>Figure 23</strong>: Participant 2, Emotional Changes in Human-Human Explanation and Solving the Task</figcaption></figure><p>Participant 2 experienced a range of emotions throughout Task 2 (Figure 24), primarily because the task was unfamiliar to her.</p><p>She frequently displayed <strong>negative emotions</strong> such as confusion, upset, and nervousness. Even after receiving AI support, these emotions persisted. However, following guidance from the moderator, she showed signs of surprise and was able to transition from negative to positive emotions. As a result, she appeared more relaxed upon successfully completing the task.</p><figure><img alt="Diagram showing Participant 2’s emotional changes during AI-human explanations while completing the task 2" src="https://cdn-images-1.medium.com/max/1024/1*ELDO4t4AASgWOXaO0UPdeg.png" /><figcaption><strong>Figure 24</strong>: Participant 2, Emotional Changes in AI-Human Explanation, and Solving the Task</figcaption></figure><h4>5.5 Participants’ post-test analysis</h4><p>Table 5 presents the differences between Participant 1 and Participant 2 in their <strong>post-test results</strong>. The initial set of questions aimed to capture the users’ feelings about the task they had just completed, and these were asked immediately after task completion to ensure accuracy and relevance. Additionally, the questions sought to assess participants’ impressions of the type of explanation they received.</p><blockquote><em>As shown in Table 5, although the AI explanation caused some confusion, Participant 1 reported greater satisfaction with it than Participant 2. Interestingly, Participant 2’s response to the question “What do you like most about this process?” was “Learning something new,” suggesting a positive aspect of the experience despite an overall negative impression of the AI.</em></blockquote><p>Similar questions were posed following the human explanation. Here, Participant 1, who had shown higher satisfaction with AI support, expressed <strong>frustration with the human explanation</strong>, which did not effectively assist in solving the task. In contrast, Participant 2 reported <strong>high satisfaction with the human explanation</strong>.</p><p>A comparison of the two participants’ results indicates that their preferences remained consistent at the end of the test.</p><blockquote><strong>Participant 1 continued to favour AI support</strong>, while Participant 2 maintained a preference for human explanation. Notably, both participants were surprised by Task 2, as it involved unfamiliar content.</blockquote><figure><img alt="Table showing post-test questions for Participants 1 and 2, recorded thier emotions, and participants’ answers as data" src="https://cdn-images-1.medium.com/max/1015/0*a71wG1jyVPBF1Q-t" /><figcaption><strong>Table 5</strong>: Post-test questions, answers from both participants, and how the author feels the emotions in each stage</figcaption></figure><h3>6. Discussion and conclusions</h3><p>This pilot study aimed to explore how human-human and human-AI explanations influence user frustration and satisfaction when completing tasks in Blackboard. While the small sample size limits generalisability, the case study provided valuable insights.</p><h4><strong>Key findings:</strong></h4><ul><li>Both participants had intermediate experience with Blackboard.</li><li>Participant 1 <strong>maintained a positive view of AI</strong>; Participant 2 consistently held a <strong>negative view</strong>.</li><li>Human support satisfaction varied: Participant 1 <strong>preferred</strong> human assistance, while Participant 2 remained <strong>dissatisfied</strong>.</li><li><strong>AI support led to mixed results</strong>, Participant 1 found it helpful despite confusion, whereas Participant 2 found it stressful and frustrating.</li></ul><blockquote>Despite limitations, circumplex models built from session recordings and post-test questionnaires revealed notable emotional patterns.</blockquote><h4>6.1 Participant 1: Expert’s analysis summary</h4><ul><li><strong>Human-AI Support</strong>: Participant 1 received the first AI message (4) during a stressful moment (Figure 25) but overlooked it (5) due to the low visibility of the Microsoft Teams chat. Mistaking the AI for the moderator (8), they took 1 minute and 12 seconds (Figure 17) to permit its assistance. Confusion arose mid-task (10) and near the end (11) from unfamiliarity with the test environment, but they ultimately expressed satisfaction (13) with the AI’s explanation.</li></ul><figure><img alt="Circumplex model diagram showing Participant 1’s emotions during each step of the human-AI interaction" src="https://cdn-images-1.medium.com/max/1024/1*YfwdCeUVrJmjv8gsJxfzng.png" /><figcaption><strong>Figure 25</strong>: Participant 1: Human-AI situation circumplex model based on the expert’s emotion analysis</figcaption></figure><ul><li><strong>Human-Human Support</strong>: Participant was upset about not completing Task 1 alone. Although they completed Task 2, the emotional response appeared subdued and possibly inauthentic. <strong>Frustration was evident toward the end.</strong></li></ul><figure><img alt="Circumplex model diagram showing Participant 1’s emotions during each step of the human-human interaction" src="https://cdn-images-1.medium.com/max/1024/1*rnHiF7BGmC1JnKuWBFJc0w.png" /><figcaption><strong>Figure 26</strong>: Participant 1: Human-human situation circumplex model based on the expert’s emotion analysis</figcaption></figure><blockquote><strong>Comparison</strong>: AI support led to faster resolution but more confusion. Human support offered less clarity. Overall, Participant 1 supported the hypothesis that AI can be effective and satisfying.</blockquote><figure><img alt="Circumplex model diagram showing Participant 1’s emotions during each step of the human-AI and human-human interaction" src="https://cdn-images-1.medium.com/max/680/1*u2GgHQmIzNQTwyxz4k_eEA.png" /><figcaption><strong>Figure 27</strong>: Participant 1: Human-human and human-AI explanation situation comparison based on the expert’s emotion analysis</figcaption></figure><h4>6.2 Participant 2: Expert’s analysis summary</h4><ul><li><strong>Human-AI Support</strong>: Faced difficulty noticing AI messages. Required moderator intervention, which affected test neutrality. Felt frustrated and panicked during the task, though eventually completed it. Expressed dissatisfaction in the post-test questionnaire.</li></ul><figure><img alt="Circumplex model diagram showing Participant 2’s emotions during each step of the human-AI interaction" src="https://cdn-images-1.medium.com/max/1024/1*bdvVcqBX8UrC4dL5wDw8ug.png" /><figcaption><strong>Figure 28</strong>: Participant 2: Human-AI situation circumplex model based on the expert’s emotion analysis</figcaption></figure><ul><li><strong>Human-Human Support</strong>: Felt comfortable and satisfied throughout, showing consistent positive emotions.</li></ul><figure><img alt="Circumplex model diagram showing Participant 2’s emotions during each step of the human-human interaction" src="https://cdn-images-1.medium.com/max/1024/1*XzE_J7m_AKRHxpER38-JZA.png" /><figcaption><strong>Figure 29</strong>: Participant 2: Human-human situation circumplex model based on the expert’s emotion analysis</figcaption></figure><blockquote><strong>Comparison</strong>: Participant 2 preferred human explanations and <strong>did not support the hypothesis.</strong></blockquote><figure><img alt="Circumplex model diagram showing Participant 2’s emotions during each step of the human-AI and human-human interaction" src="https://cdn-images-1.medium.com/max/680/1*Rlbqw0sG6Nw2SnrjljId8Q.png" /><figcaption><strong>Figure 30</strong>: Participant 2: Human-human and human-AI explanation situation comparison based on the expert’s emotion analysis</figcaption></figure><h4>6.3 Both participants: Expert’s analysis summary</h4><p>Both participants experienced more <strong>frustration with AI support</strong>. However, Participant 1 attributed this to interface confusion, not AI quality. Participant 2’s frustration stemmed from a lack of trust in AI. The different AI sources (ChatGPT vs. Blackboard Help Center) may have influenced emotional responses.</p><figure><img alt="Circumplex model diagram showing both Participant’s emotions during each step of the human-AI and human-human interaction" src="https://cdn-images-1.medium.com/max/1024/1*fDfQ6Zwr37uvWEevbIuPuA.png" /><figcaption><strong>Figure 31</strong>: Both participants’ human-human and human-AI explanation situation comparisons based on the expert’s emotion analysis</figcaption></figure><h4>6.4 Final conclusion</h4><p>The study suggests:</p><ul><li>Participants’ attitudes toward AI impact their experience.</li><li>AI support can be quicker but may not always reduce frustration.</li><li>Human support offers emotional comfort, though not always technical efficiency.</li></ul><p>While the results are inconclusive due to a small sample size and other limitations, they offer insights for future work on AI-based support systems in educational platforms like Blackboard. The study reinforces previous research on gender differences in emotion recognition and highlights the importance of tailoring AI responses accordingly.</p><h4>6.5 Project reflection</h4><p>After successfully conducting this study with relevant and accurate data, a proposed chat agent (Figure 32) will be implemented for Blackboard to assist students in need of technical support.</p><blockquote>The chat agent will be activated when a user experiences confusion with the interface.</blockquote><blockquote><strong>The findings of this study indicate that male and female facial expressions are different</strong>, so the proposed chat agent will not only identify frustration emotions on a user’s face but also detect if the user is stuck in the same place or experiencing unnecessary clicks.</blockquote><figure><img alt="Diagram showing a new GCU student interacting with a blackboard, capturing emotional changes, and activating AI support" src="https://cdn-images-1.medium.com/max/1024/1*fveHr5BEtVFPa1HkrIz46g.png" /><figcaption><strong>Figure 32</strong>: Proposed AI chat agent on the blackboard</figcaption></figure><h3>7. Future work</h3><p>Given the exploratory nature and limited sample size of this pilot study, several opportunities exist for future research:</p><ul><li><strong>Larger Sample Size</strong>: Future studies should involve a more diverse and statistically significant number of participants to strengthen the validity and generalisability of findings.</li><li><strong>Consistent AI Support Source</strong>: The use of two different AI sources (ChatGPT and Blackboard Help Centre) may have influenced participant reactions. A controlled, consistent AI source should be used in future experiments to ensure fair comparisons.</li><li><strong>Balanced Task Assignment</strong>: Each participant should receive AI and human support in both familiar and unfamiliar tasks to assess frustration and satisfaction under varied conditions, using a proper counterbalanced design.</li><li><strong>Multiple Human Supporters</strong>: Using more than one caller or human assistant will help avoid potential bias and allow for better counterbalancing between participants.</li><li><strong>Advanced Emotion Recognition Tools</strong>: Implementing software such as iMotions or Affectiva can improve the accuracy of emotional analysis, especially in detecting subtle expressions that may not be visually evident, particularly in male participants.</li><li><strong>Custom-Built AI Interface</strong>: Future research could benefit from developing a purpose-built AI support interface tailored to the Blackboard environment, rather than relying on existing chat tools like Microsoft Teams, which may obscure important messages.</li><li><strong>Longitudinal Study</strong>: Tracking participants over time could reveal how perceptions of AI and human support evolve with continued exposure and experience.</li><li><strong>Improved Facial Emotion Recognition Models</strong>: Given gender-based variation in facial expressiveness, future AI systems should be trained with more inclusive datasets to improve emotional detection accuracy across different user groups.</li></ul><p>Collectively, these improvements will enhance the reliability of the findings and inform the design of more adaptive, emotionally aware support systems for digital learning environments.</p><h3>References</h3><p><em>Activation-integration model coefficients for Biopsy</em> (2022) [Screenshot]. Available at: <a href="https://doi.org/10.1145/3514094.3534150">https://doi.org/10.1145/3514094.3534150</a> (Accessed: 17 February 2023).</p><p>Avula, S., Chadwick, G., Arguello, J. and Capra, R. (2018) ‘SearchBots: User Engagement with ChatBots during Collaborative Search’, <em>CHIIR ’18: Proceedings of the 2018 Conference on Human Information Interaction &amp; Retrieval, </em>New Brunswick, NJ, USA, March 2018. Available at: <a href="https://doi.org/10.1145/3176349.3176380">https://doi.org/10.1145/3176349.3176380</a> (Accessed: 5 April 2023)</p><p><em>Blogs</em> (2023) [Screenshot]. Available at: <a href="https://help.blackboard.com/Learn/Student/Original/Interact/Blogs">https://help.blackboard.com/Learn/Student/Original/Interact/Blogs</a> (Accessed: 28 March 2023).</p><p><em>ChatGPT</em> (2023) [Screenshot]. Available at: <a href="https://chat.openai.com/">https://chat.openai.com</a> (Accessed: 28 March 2023).</p><p>Chaves, A. P. and Gerosa, M. A. (2018) ‘Single or Multiple Conversational Agents? An Interactional Coherence Comparison’, <em>CHI ’18: Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems</em>, Montreal QC, Canada, April 2018. Available at: <a href="https://doi.org/10.1145/3173574.3173765">https://doi.org/10.1145/3173574.3173765</a> (Accessed: 5 April 2023)</p><p>Chacko, A. (2023) <em>The complete guide to chatbots for marketing. </em>Available at: <a href="https://sproutsocial.com/insights/chatbot-marketing">https://sproutsocial.com/insights/chatbot-marketing</a> (Accessed: 5 May 2023)</p><p>Dumas, J. and Fox, J. (2007) <em>Usability testing: Current practice and future directions</em>, 2nd edn. New York: Lawrence Erlbaum Associates. Available at: <a href="https://doi.org/10.1201/9781410615862">https://doi.org/10.1201/9781410615862</a> (Accessed: 29 April 2023).</p><p>Duijst, D. (2017) ‘Can we improve the user experience of chatbots with personalisation’, <em>Master’s thesis</em>. University of Amsterdam. Available at: <a href="https://www.researchgate.net/profile/Danielle-Duijst/publication/318404775_Can_we_Improve_the_User_Experience_of_Chatbots_with_Personalisation/links/5967ba16a6fdcc18ea662ce7/Can-we-Improve-the-User-Experience-of-Chatbots-with-Personalisation.pdf">https://www.researchgate.net/profile/Danielle-Duijst/publication/318404775_Can_we_Improve_the_User_Experience_of_Chatbots_with_Personalisation/links/5967ba16a6fdcc18ea662ce7/Can-we-Improve-the-User-Experience-of-Chatbots-with-Personalisation.pdf</a> (Accessed: 5 May 2023).</p><p>Fan, M., Shi, S. and Truong, K. (2020) ‘Practices and challenges of using think-aloud protocols in industry: an international survey’, <em>Journal of Usability Studies</em>, 15(2), pp. 85–102. Available at: <a href="https://dl.acm.org/doi/abs/10.5555/3532708.3532711">https://dl.acm.org/doi/abs/10.5555/3532708.3532711</a> (Accessed: 17 February 2023).</p><p>Fan, M., Yang, X., Yu, T., Liao, V. and Zhao, J. (2022) ‘Human-AI Collaboration for UX Evaluation: Effects of Explanation and Synchronization’, <em>Proceedings of the ACM on Human-Computer Interaction,</em> 6, article number 96. Available at: <a href="https://doi.org/10.1145/3512943">https://doi.org/10.1145/3512943</a> (Accessed: 17 February 2023).</p><p>Gnewuch, U., Morana, S. and Maedche, A. (2017) ‘Towards Designing Cooperative and Social Conversational Agents for Customer Service’, <em>Proceedings of the International Conference on Information Systems (ICIS)</em> <em>2017</em>. Available at: <a href="https://www.researchgate.net/profile/Ulrich-Gnewuch/publication/320015931_Towards_Designing_Cooperative_and_Social_Conversational_Agents_for_Customer_Service/links/59c8d1220f7e9bd2c01a38a5/Towards-Designing-Cooperative-and-Social-Conversational-Agents-for-Customer-Service.pdf">https://www.researchgate.net/profile/Ulrich-Gnewuch/publication/320015931_Towards_Designing_Cooperative_and_Social_Conversational_Agents_for_Customer_Service/links/59c8d1220f7e9bd2c01a38a5/Towards-Designing-Cooperative-and-Social-Conversational-Agents-for-Customer-Service.pdf</a> (Accessed: 5 May 2023).</p><p>Hayashi, Y., and Wakabayashi, K. (2017) ‘Can AI become Reliable Source to Support Human Decision Making in a Court Scene?’,<em> Proceedings of the ACM on Human-Computer Interaction</em>, 4, pp. 1–22. Available at: <a href="https://doi.org/10.1145/3415173">https://doi.org/10.1145/3415173</a> (Accessed: 17 February 2023).</p><p><em>Introducing ChatGPT </em>(2023) Available at: <a href="https://openai.com/blog/chatgpt">https://openai.com/blog/chatgpt</a> (Accessed: 2 March 2023).</p><p>Jain, M., Kumar, P., Kota, R. and Patel, S. N. (2018) ‘Evaluating and Informing the Design of Chatbots’, <em>DIS ’18: Proceedings of the 2018 Designing Interactive Systems Conference</em>, Hong Kong, China, June 2018, pp 895–906. Available at: <a href="https://doi.org/10.1145/3196709.3196735">https://doi.org/10.1145/3196709.3196735</a> (Accessed: 5 May 2023)</p><p>Kou, Y. and Gui, X. (2020) ‘Mediating Community-AI Interaction through Situated Explanation: The Case of AI-Led Moderation’, <em>Proceedings of the ACM on Human-Computer Interaction</em>, 4, pp. 1–22. Available at: <a href="https://doi.org/10.1145/3415173">https://doi.org/10.1145/3415173</a> (Accessed: 17 February 2023).</p><p>Kuniavsky, M. (2003) <em>Observing the user experience: A practitioner’s guide to user research</em>. San Francisco: Morgan Kaufmann publishers.</p><p>Kwon, J., Ha, J., Kim,-H., Choi, J. W. and Kim, L. (2021) ‘Emotion Recognition Using a Glasses-Type Wearable Device via Multi-Channel Facial Responses’, <em>IEEE Access</em>, 9, pp. 146392–146403. Available at: <a href="https://doi.org/10.1109/ACCESS.2021.3121543">https://doi.org/10.1109/ACCESS.2021.3121543</a> (Accessed: 24 March 2023).</p><p>Lazar, J., Feng, J.H. and Hochheiser, H. (2009) <em>Research methods in human computer interaction. </em>5th edn. United Kingdom: John Wiley &amp; Songs Ltd.</p><p>O’Connor, S. and ChatGPT (2023) ‘Open artificial intelligence platforms in nursing education: Tools for academic progress or abuse?’, <em>Nurse Education in Practice</em>, 66. Available at: <a href="https://doi.org/10.1016/j.nepr.2022.103537">https://doi.org/10.1016/j.nepr.2022.103537</a> (Accessed: 21 March 2023)</p><p>Porcheron, M., Fischer, J. E. and Reeves, S. (2021) ‘Pulling Back the Curtain on the Wizards of Oz’, <em>Proceedings of the ACM on Human-Computer Interaction</em>, 4, article number 243. Available at: <a href="https://doi.org/10.1145/3432942">https://doi.org/10.1145/3432942</a> (Accessed: 27 March 2023)</p><p>Quesenbery, W. (2014)<em> </em>‘Balancing the 5Es: Usability’, <em>Cutter IT Journal</em>, 17(2), pp. 5. Available at: <a href="https://wqusability.com/articles/5es-citj0204.pdf">https://wqusability.com/articles/5es-citj0204.pdf</a> (Accessed: 22 March 2023)</p><p>Ramaswamy, S. (2023) <em>The Wizard of Oz Method in UX</em>. Available at: <a href="https://www.nngroup.com/articles/wizard-of-oz/">https://www.nngroup.com/articles/wizard-of-oz/</a> (Accessed: 27 March 2023)</p><p>Russell, J. A. (1980) ‘A circumplex model of affect’,<em> Journal of Personality and Social Psychology</em>, 39(6), pp. 1161–1178. Available at: <a href="https://doi.org/10.1037/h0077714">https://doi.org/10.1037/h0077714</a> (Accessed: 2 April 2023)</p><p>Silvervarg, A., and Jönsson, A. (2013) ‘Iterative development and evaluation of a social conversational agent’, In<em> Proceedings of the Sixth International Joint Conference on Natural Language Processing,</em> pp. 1223–1229. Available at: <a href="https://aclanthology.org/I13-1174.pdf">https://aclanthology.org/I13-1174.pdf</a> (Accessed: 2 April 2023)</p><p><em>The user interface of AI Assistant that presents the AI-suggested problems synchronously with explanations</em> (2022) [Screenshot]. Available at: <a href="https://doi.org/10.1145/3512943">https://doi.org/10.1145/3512943</a> (Accessed: 17 February 2023)</p><p>Vodrahalli, K., Daneshjou, R., Gerstenberg, T. and Zou, J. (2022) ‘Do Humans Trust Advice More If It Comes from AI? An Analysis of Human-AI Interactions’, <em>Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and Society, </em>New York, NY, USA, July 2022. Available at: <a href="https://doi.org/10.1145/3514094.3534150">https://doi.org/10.1145/3514094.3534150</a> (Accessed: 17 February 2023)</p><p><em>Visualization of our study design</em> (2022) [Screenshot]. Available at: <a href="https://doi.org/10.1145/3514094.3534150">https://doi.org/10.1145/3514094.3534150</a> (Accessed: 17 February 2023)</p><p>Valério, F. A., Guimarães, T. G., Prates, R. O., and Candello, H. (2017) ‘Here’s What I Can Do: Chatbots’ Strategies to Convey Their Features to Users’, <em>IHC 2017: Proceedings of the XVI Brazilian Symposium on Human Factors in Computing Systems,</em> Joinville, Brazil, pp. 1–10. Available at: <a href="https://doi.org/10.1145/3160504.3160544">https://doi.org/10.1145/3160504.3160544</a> (Accessed: 5 May 2023)</p><p>Wixon, D. (2003) ‘Evaluating usability methods: Why the current literature fails the practitioner’, <em>Interactions, </em>10(4), pp. 28–34. Available at: <a href="https://doi.org/10.1145/838830.838870">https://doi.org/10.1145/838830.838870</a> (Accessed: 20 April 2023)</p><p>Wang, D., Weisz, J., Muller, M., Ram, P., Geyer, M., Dugan, C., Tausczik, Y., Samulowitz, H. and Gray, A. (2019) ‘Human-AI Collaboration in Data Science: Exploring Data Scientists’ Perceptions of Automated AI’, <em>Proceedings of the ACM on Human-Computer Interaction</em>, 3, article number 211. Available at: <a href="https://doi.org/10.1145/3359313">https://doi.org/10.1145/3359313</a> (Accessed: 17 March 2023)</p><h4>Appendix A</h4><figure><img alt="Diagram showing how participants completed Task 1 and Task 2, including their actions and task flow" src="https://cdn-images-1.medium.com/max/1024/1*kY9krYjQry80xZULKbEtGg.png" /><figcaption><strong>Figure 33: </strong>Pilot test — Task 1 and 2</figcaption></figure><h4>Appendix B</h4><figure><img alt="Data table showing problems, methods used, prompts, and ChatGPT responses" src="https://cdn-images-1.medium.com/max/1024/1*NycHPoKGOmTXVr1t5yW8lg.png" /><figcaption><strong>Figure 34</strong>: Document recent issues encountered by new Blackboard users and have GPT-3 respond to solve those issues [<a href="https://docs.google.com/spreadsheets/d/1omzbT0RGldzl-imIB_CpAPyruhNTDnk0dmoVpQEY9Lk/edit#gid=0">Link</a>].</figcaption></figure><h4>Appendix C</h4><figure><img alt="Pilot test checklist showing actions and corresponding times in sequence" src="https://cdn-images-1.medium.com/max/512/1*XENPnqQJLXy_1qUIP40ntg.png" /><figcaption><strong>Figure 35</strong>: Pilot test — Check list</figcaption></figure><h4>Appendix D</h4><figure><img alt="Six pre-test questions asked by the caller" src="https://cdn-images-1.medium.com/max/1024/1*QXnBJeVIx_HxY186B4dFKQ.png" /><figcaption><strong>Figure 36</strong>: Pre-test Questionnaire — caller</figcaption></figure><h4>Appendix E</h4><figure><img alt="Nine pre-test questions asked by the both participants" src="https://cdn-images-1.medium.com/max/1024/1*JuEtG1aGcxQJLB0cXAsK1g.png" /><figcaption><strong>Figure 37</strong>: Pre-test Questionnaire — Participant 1/2</figcaption></figure><h4>Appendix F</h4><figure><img alt="Post-test interview questions asked to both participants about their experiences with AI explanations, human explanations, and overall task experience" src="https://cdn-images-1.medium.com/max/1024/1*KKepe2tsWbc0uOLrHxKXuw.png" /><figcaption><strong>Figure 38</strong>: Post test interview — Participant 1 and 2</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aefb6e985362" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/human-or-ai-measuring-the-impact-of-explanation-style-in-an-educational-support-system-aefb6e985362">Human or AI? Measuring the impact of explanation style in an educational support system</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Should UX Designers Trust AI with User Behaviour Analysis?]]></title>
            <link>https://medium.com/design-bootcamp/should-ux-designers-trust-ai-with-user-behaviour-analysis-53c9ec629f9c?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/53c9ec629f9c</guid>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[user-research]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Tue, 22 Jul 2025 15:59:14 GMT</pubDate>
            <atom:updated>2025-08-18T06:55:24.080Z</atom:updated>
            <content:encoded><![CDATA[<h4>Lessons from Using Generative AI for Heatmaps and Design Decisions</h4><p>The main reason for writing this blog is that one of my mentees was asked by his team lead to use AI to analyse user behaviour. I wanted to demonstrate how generative AI can act as a valuable team member in the UX process, supporting our work rather than replacing human judgement, especially in a world that often values quantity over quality. For example, some agencies might want numerous user behaviour reports generated quickly using AI, <strong>without checking whether those reports are actually accurate or useful for real design decisions</strong>.</p><figure><img alt="AI helping a designer with their design work in an office" src="https://cdn-images-1.medium.com/max/1024/1*xsmLu3CaLLbj8eKnatsyfA.jpeg" /><figcaption><strong>Figure 1</strong>: Human AI interaction</figcaption></figure><h4><strong>Problem</strong></h4><p>In many UX teams, there is growing pressure to produce numerous AI-generated user behaviour reports quickly. While this satisfies the demand for speed and volume, it often risks overlooking the accuracy, depth, and practical value of the insights. This trend can lead to design decisions based on incomplete or misleading information, especially when AI outputs are accepted without human review.</p><h4><strong>Methods</strong></h4><ul><li>Used AI-powered <strong>Microsoft Clarity’s “Summarise heatmaps”</strong> feature to analyse user interactions.</li><li>Ran <strong>prompt-based queries</strong> in <strong>ChatGPT</strong> and <strong>Perplexity AI</strong> to gather UX-related insights and recommendations.</li><li>Compared <strong>AI-generated outputs</strong> with <strong>human interpretation</strong> to assess accuracy and relevance.</li></ul><h4><strong>Improvements</strong></h4><ul><li>Encourage the use of <strong>multiple AI tools</strong> and <strong>iterative prompt testing</strong> to refine the quality of outputs.</li><li>Always apply <strong>UX expertise</strong> to verify AI insights, ensuring they are relevant, actionable, and aligned with project goals.</li><li>Treat AI as a <strong>collaborative assistant</strong>, not a replacement for human analysis.</li></ul><h4><strong>Results</strong></h4><ul><li>AI tools delivered <strong>fast summaries and ideas</strong>, reducing time spent on initial data exploration.</li><li>Outputs were sometimes <strong>too general</strong> or missed specific behaviour patterns (e.g., clicks, scroll depth, attention focus).</li><li>ChatGPT and Perplexity AI were effective in <strong>reminding designers of principles</strong> and <strong>suggesting ideas</strong>, but occasionally produced irrelevant or inaccurate recommendations.</li><li>Combining AI speed with human critical thinking resulted in <strong>more accurate and useful behaviour insights</strong> for UX decisions.</li></ul><h3>Do generative AI tools really help with UX work?</h3><p>The simple answer is <strong>YES</strong>, but with important <strong>LIMITATIONS</strong>. AI tools can support UX designers in various ways, such as <strong>predicting user behaviour</strong>, <strong>personalising experiences</strong>, and <strong>streamlining prototyping </strong>(Soegaard, 2025). Liu, Zhang, and Budiu (2023) also highlight how generative AI bots assist UX professionals as content editors, research assistants, brainstorming partners, and design helpers.</p><blockquote>However, if you believe you can do all design work without any human design knowledge by relying solely on AI, I would love to hear about your end-user experience.</blockquote><p>Some argue AI tools might replace designers, researchers, or even end users (Liu and Moran, 2023). Yet, we <strong>must be cautious and not trust everything AI generates because all tools have limitations</strong>. It’s crucial to request different versions, try various tools and prompts, and apply UX knowledge to evaluate the accuracy of AI outputs. <strong>Not all AI-generated content is accurate or reliable</strong>.</p><p>With this mindset, I started experimenting to see how AI can support me as a teammate in data analysis and how it fits into collaborative settings.</p><figure><img alt="AI assisting a UI/UX designer with their design work at home during the night." src="https://cdn-images-1.medium.com/max/1024/1*gZ8MDoSzeYzSkx9Q1OK8ww.jpeg" /><figcaption><strong>Figure 2</strong>: Do generative AI tools really help with UX work?</figcaption></figure><h3>My experience using AI in UX design</h3><p>As a UI/UX designer and web developer, AI tools have supported me multiple times a day. I use generative AI like ChatGPT, Midjourney, and Adobe Firefly, just as many designers do.</p><p>However, if any company thinks AI can completely replace human roles right now, that mindset may be premature.</p><p>By the end of this blog, you will understand why. AI is not yet fully trained to grasp every nuance of human behaviour or context, but it can certainly serve as a valuable team member (assisting, suggesting, and analysing) while <strong>final decisions remain human-led </strong>(Liu and Moran, 2023).</p><h3>What you will find in this blog</h3><ul><li>Experiments conducted using AI for user behaviour analytics</li><li>Heatmap analysis using Microsoft Clarity AI</li><li>Prompt-based insights from ChatGPT and Perplexity AI</li><li>My reflections and analysis</li><li>How AI suggestions combined with personal expertise improved results</li></ul><h3>Experiment 1: Heatmap Analysis with Microsoft Clarity AI</h3><p>I have previously explained how I boosted conversions using MS Clarity heatmap and video behaviour analysis in <a href="https://ruwanthisulanjali.medium.com/how-ux-design-increased-a-law-firms-homepage-conversions-by-400-2f05b48649ad"><strong>this UX case study</strong></a>.</p><p>For this new project, I wanted to experiment further with AI. Earlier attempts were ineffective, so this time I used a different approach.</p><p>In the MS Clarity heatmap interface, I simply clicked “Summarise heatmaps” (Figure 3), and within minutes, I received an analysis of user behaviour and key takeaways. While the process was very fast, the question remained: how useful was this analysis? (Figures 5 and 6).</p><blockquote>It’s important to remember to carefully review AI-generated content to verify its accuracy.</blockquote><blockquote>MS Clarity itself cautions that “mistakes are possible.”</blockquote><figure><img alt="AI-generated insights from MS Clarity heatmaps and session recordings." src="https://cdn-images-1.medium.com/max/1024/1*kUjKzDiSPZit74LlF1lCjw.png" /><figcaption><strong>Figure 3:</strong> MS Clarity heatmaps insights and recording insights by AI</figcaption></figure><blockquote>My first impression was that the analysis was too shallow.</blockquote><p>Usually, I note many detailed insights when analysing heatmaps, but this AI-generated data only provided a rough overview.</p><blockquote><strong>Limitation:</strong> If I focus only on homepage PC clicks and use MS Clarity AI, it generates the results as a single summary (Figure 3), providing just one point each for mobile, tablet, and desktop. It does not provide separate heatmap analyses for clicks, scrolls, or attention.</blockquote><blockquote>However, ChatGPT and Perplexity can help with separate analyses (Figures 5 and 6).</blockquote><p>After documenting this data, I uploaded the heatmap images into ChatGPT Plus (which supports multiple images) and perplexity for further iterations. Later, I compared all insights with my own analysis and made improvements (Figures 5 and 6).</p><h3>Experiment 2: Using ChatGPT and Perplexity AI for Analytical Summaries</h3><p>Next, I used ChatGPT and perplexity to analyse heatmaps, experimenting with different prompts to elicit the level of detail needed for the next iteration of design work. Moran (2024) advises that effective outcomes from generative AI require thoughtfully crafted prompts that provide context, specify the task, outline guidelines, and include examples.</p><figure><img alt="GPT interface for data analysis and reporting." src="https://cdn-images-1.medium.com/max/1024/1*Ru9TNlN4SwpLScM3YFYXyQ.png" /><figcaption><strong>Figure 4:</strong> Data analysis and report GPT I used here</figcaption></figure><p>In Chat GPT, I chose the “Data Analysis and Report AI” GPT variant because it delivers more detailed insights than the standard chat interface.</p><p>Here’s a recent prompt that worked fairly well (though not perfectly):</p><pre><strong>Context</strong>:<br>I&#39;m a UX designer working on a CRO project.<br>We&#39;re updating the <em>[X]</em> client homepage <em>[Home page link]</em> to improve conversions.<br><br><strong>Ask</strong>:<br>You&#39;re a UX analyst / UX researcher.<br>1. Analyse the attached MS Clarity heatmaps.<br>2. Describe what the user behaviour looks like.<br>3. Share what you can see that is good or bad.<br>4. Suggest improvements.<br><br><strong>Rules</strong>:<br>1. Tone: clear, approachable, and action-oriented<br>2. Consider design principles, guidelines, best practices<br>(e.g. Shneiderman&#39;s 8 golden rules, Nielsen Norman heuristics,<br>usability goals, and visual design principles)<br>3. Consider design trends and competitor opportunities<br>4. Avoid jargon and superlatives<br><br><strong>Examples</strong>:<br>Not included**<br><br><br>**In this experiment, I didn&#39;t add examples because sometimes if I add<br>examples, it only gives data based on that area, and I didn&#39;t want it to be <br>fixed in a box. In the &quot;Rules&quot; section, I listed the design principles and <br>guidelines clearly. But if you like, you can try adding examples. <br>It didn&#39;t work earlier for me, so I didn&#39;t include them here.</pre><blockquote>If you have better prompt suggestions, I’m happy to experiment further.</blockquote><p>The speed of AI responses is impressive, but the key question remains: <strong>how reliable are these insights?</strong></p><p>After a quick review, I found some points questionable when cross-checked against my own knowledge and observations. Some recommendations were helpful, while others missed the mark (Figure 5 and 6).</p><h3>Experiment 3: My ( human ) analysis and key takeaways for the final output</h3><p>After gathering AI-generated data from MS Clarity, ChatGPT, and Perplexity, I conducted my own analysis to explore how deeply I could redesign the page.</p><p>I performed two analyses:</p><ul><li>Heatmap data (clicks, scrolls, and attention analysis)</li><li>Session recordings to observe real user behaviour and identify confusion points</li></ul><p>I especially value session recordings because they reveal where users get stuck or frustrated.</p><blockquote><strong>Note: </strong><em>This section focuses solely on analysing homepage PC clicks. Although analyses were conducted for PC, mobile, tablet, scroll behaviour, and attention patterns, listing every finding here would make the blog too long.</em></blockquote><h4>Home page clicks analysis [PC]</h4><figure><img alt="MS Clarity heatmap screnshots of homepage clicks, including insights from Human analysis and Generative AI tools (ChatGPT, MS Clarity, and Perplexity), highlighting key findings." src="https://cdn-images-1.medium.com/max/1024/1*mGvZ7pX16O5bexGN4bvSvA.png" /><figcaption><strong>Figure 5: </strong>Homepage clicks [PC] — Human and AI analysis for heatmap insights</figcaption></figure><h4>Session recording analysis</h4><p>I reviewed over 10 session recordings for human analysis and used MS Clarity AI to analyse the top 10 videos.</p><blockquote><strong><em>Limitation</em></strong><em>:</em> I tried uploading one video to Perplexity AI but had to reduce its size due to file limits, a lengthy process if done for many videos. Thus, I relied on AI for quicker outputs but depended on human-led insights for deeper understanding.</blockquote><p>Here, you can see how human and AI analyses for video behavior analysis complement each other, leading to better-informed design decisions.</p><figure><img alt="Behaviour analysis of MS Clarity session recordings on the client’s website, with key findings from both human analysis and Generative AI insights." src="https://cdn-images-1.medium.com/max/1024/1*Vb0nzb_sUvxFIYCGq2floA.png" /><figcaption><strong>Figure 6: </strong>Video behaviour analysis — Human and AI session recording insights</figcaption></figure><h3>Key Takeaways</h3><p>As many experts have noted, <strong>AI has clear limitations</strong>. It is a valuable team member, not a replacement. Based on my experience, we cannot fully rely on AI for detailed user behaviour analysis such as heatmaps and session recordings.</p><p>For example, some ChatGPT and most Perplexity AI heatmap analyses <strong>contained errors</strong>. If redesigns were based solely on these, we could see no improvement, or worse, performance could decline.</p><blockquote>This is where human insight is crucial.</blockquote><blockquote>On the flip side, Perplexity AI surfaced some points I had missed. When overwhelmed with data, even experienced designers can overlook principles and guidelines, at those moments, AI can serve as a helpful teammate, highlighting blind spots.</blockquote><p>Another major limitation is that current generative AI tools cannot access external links directly (including MS Clarity URLs). We must download videos or export session data, then upload files within AI size limits, which is time-consuming. MS Clarity’s AI analysis was helpful but often generic, without the ability to add prompts. Improvements are likely in the future, but for now, it speeds up a time-intensive process.</p><p><strong>For lead-generation user journeys, combining Google Analytics with session recordings remains mostly a human-led</strong> task based on my experience. This is mainly because MS Clarity’s video behaviour analysis doesn’t capture all the necessary details, and it doesn’t allow adding text prompts for specific insights. ChatGPT and Perplexity also have limitations with uploading videos, so user journey analysis is still largely human-led. This approach provides a clearer understanding of which pages need to be redesigned first, rather than wasting time redesigning every page without knowing their actual impact.</p><p>Suggestions on improving this workflow are welcome.</p><figure><img alt="Gen AI is helping the UX team with their design projects." src="https://cdn-images-1.medium.com/max/1024/1*o3f7Dt8VvvDCCGdNNl0xyQ.jpeg" /><figcaption><strong>Figure 7: </strong>AI is a valuable team member <strong>but not a replacement</strong>.</figcaption></figure><h3>Conclusion</h3><blockquote>Generative AI has proven itself to be a valuable team member in UX work rather than a full replacement for human expertise.</blockquote><p>These experiments showed that while AI tools like Microsoft Clarity AI, ChatGPT, and Perplexity can speed up data analysis and provide fresh perspectives, their outputs often lack the depth and contextual understanding needed for confident design decisions.</p><p>Heatmap summaries from <strong>MS Clarity AI</strong> were fast but generic, missing the nuanced behavioural insights that human analysis could uncover. It also has the <strong>limitation</strong> of not allowing text prompts to request more detailed data.</p><p><strong>ChatGPT</strong> and <strong>Perplexity</strong> offered quick suggestions, but their recommendations were <strong>sometimes incorrect or irrelevant</strong>, highlighting the risks of fully relying on AI without human evaluation. Yet, they often reminded me of design principles or insights I might overlook when overwhelmed by data, proving their strength as supportive brainstorming partners. However, they also have <strong>limitations</strong>, such as being unable to analyse video-based user behaviour or lead generation user flows.</p><p>Ultimately, AI tools are most effective when treated as <strong>collaborative team members</strong>, <strong>assisting with analysis</strong>, <strong>suggesting ideas</strong>, and <strong>speeding up repetitive tasks</strong>, while <strong>critical decisions remain human-led</strong>. Combining AI-generated insights with personal expertise ensures designs are grounded in real user behaviour and context rather than assumptions or incomplete automated summaries.</p><blockquote>Instead of replacing designers or researchers, AI can empower teams to work smarter and iterate faster, as long as its limitations are acknowledged and its outputs are critically evaluated before implementation.</blockquote><h3>Reference</h3><p>Liu, F. and Moran, K. (2023) <em>AI-Powered Tools for UX Research: Issues and Limitations. </em>Available at: <a href="https://www.nngroup.com/articles/ai-powered-tools-limitations/">https://www.nngroup.com/articles/ai-powered-tools-limitations/</a> (Accessed: 7 July 2025).</p><p>Liu, F., Zhang, M. and Budiu, R. (2023) <em>AI as a UX Assistant. </em>Available at: <a href="https://www.nngroup.com/articles/ai-roles-ux/">https://www.nngroup.com/articles/ai-roles-ux/</a> (Accessed: 7 July 2025).</p><p>Moran, K. (2024)<em> CARE: Structure for Crafting AI Prompts. </em>Available at: <a href="https://www.nngroup.com/articles/careful-prompts/">https://www.nngroup.com/articles/careful-prompts/</a> (Accessed: 15 July 2025).</p><p>Soegaard, M. (2025)<em> 8 Best AI Tools for UX Designers. </em>Available at: <a href="https://www.interaction-design.org/literature/article/ai-tools-for-ux-designers">https://www.interaction-design.org/literature/article/ai-tools-for-ux-designers</a> (Accessed: 7 July 2025).</p><h3><strong>About the author</strong></h3><p>I’m Ruwanthi (Ru) Somatilaka, a UX designer and AI enthusiast passionate about blending human-centred design with emerging technologies. With experience across product-led tech companies, I explore how AI can support better user behaviour analysis and design decisions.</p><p>This article is part of my ongoing journey into AI and UX. Follow for more insights.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=53c9ec629f9c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/should-ux-designers-trust-ai-with-user-behaviour-analysis-53c9ec629f9c">Should UX Designers Trust AI with User Behaviour Analysis?</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[12 Months, Countless Lessons: My First Year at MLT Digital]]></title>
            <link>https://ruwanthisulanjali.medium.com/12-months-countless-lessons-my-first-year-at-mlt-digital-2f995fa06245?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f995fa06245</guid>
            <category><![CDATA[ui-ux-design]]></category>
            <category><![CDATA[work-anniversary]]></category>
            <category><![CDATA[uk]]></category>
            <category><![CDATA[work]]></category>
            <category><![CDATA[front-end-development]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Thu, 29 May 2025 13:07:51 GMT</pubDate>
            <atom:updated>2025-05-29T13:09:15.170Z</atom:updated>
            <content:encoded><![CDATA[<h4>A milestone moment as a UI/UX Designer and Front-End Dev in the UK</h4><p>Staring out the window, tea in hand, I watch an airplane trace its path across the clear spring sky. Suddenly, my mind often drifts back to the day I first arrived in the Scotland, UK as an international student in 2022, filled with hope and nerves. Fast forward to April 2025, and I found myself celebrating a quiet but meaningful milestone: <strong>my first anniversary</strong> as a UI/UX designer and front-end developer at MLT Digital.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*su3tPytzS_SSYjg57FsWSQ.png" /><figcaption>MLT Digital Team in 2024</figcaption></figure><p>The road here hasn’t been smooth or linear. It’s been a mix of late nights, learning curves, small wins, and the kind of challenges that shape me in unexpected ways. But it’s also been a journey of growth, both personal and professional.</p><p>So, let me take you back to where it all began.</p><h3>Overview</h3><ol><li>The Journey Begins (2022)</li><li>MLT: From Day One to Year One</li><li>Key Highlights from the Year (2024–2025)</li><li>Lessons Learned</li><li>Final Thoughts: Gratitude, Growth, and the Road Ahead</li></ol><h3>1. The Journey Begins</h3><p>It was a rainy afternoon in September 2022 when I stepped off the plane and into a new chapter of my life. I had left behind 7 years of experience and a role at one of the best product-led tech companies back home, to start fresh in the UK as a master’s student in <em>User Experience and Interaction Design (MSc UXID)</em>.</p><p>It wasn’t an easy decision. I had a stable job, a good life, and the comfort of the familiar. But I had goals waiting patiently on a list, and I knew it was time to chase them, one by one.</p><p>Leaving everything behind wasn’t just a move to another country; it felt like stepping into the unknown, with hope in one hand and uncertainty in the other, trusting that something meaningful would unfold.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*efffalsRDFCXCS-SnP0fWQ.jpeg" /><figcaption>This quote I noticed in the Lucifer TV series is something I truly believe in, and it’s reflected in my many decisions [<a href="https://www.magicalquote.com/seriesquotes/you-have-to-risk-falling-if-you-wanna-fly/#google_vignette">Image</a>]</figcaption></figure><p>That chapter of student life was filled with its lessons, challenges, and small joys, something I’ve shared in a previous blog. But this story isn’t just about classrooms and coursework. It’s about what came next, the shift from student to skilled worker.</p><blockquote>I wrote about my student experience in a previous blog: <a href="https://ruwanthisulanjali.medium.com/international-student-adventures-my-msc-in-user-experience-and-interaction-design-uxid-journey-7560d3660005">My MSc in User Experience and Interaction Design (UXID) Journey</a>.</blockquote><h4>The Struggle…</h4><p>Like I said earlier, it wasn’t easy. I felt like I was falling financially (after leaving a good company with a good salary) and mentally (because I had to take care of my parents and myself, all while being away from loved ones).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rFqhKrDjzegIOlKh2X3m3Q.jpeg" /><figcaption>Photo by <a href="https://www.pexels.com/photo/woman-sitting-on-chair-while-leaning-on-laptop-3791136/">Andrea Piacquadio</a></figcaption></figure><p>From day one in the UK, the struggle was real. I searched for UI/UX roles throughout my master’s course. Despite my experience, rejection became a familiar word. However, I was get only 2 interviews with well-known companies, and both ended quickly because of student work hour limits. During term time, I was only allowed to work 20 hours a week. That was a wall I couldn’t get past.</p><h4>Never Give Up…</h4><p>I don’t give up easily. Many people around me suggested caregiving jobs as the easiest path to sponsorship. Anyway, I don’t blame anyone for that advice, after all, their goal was simple: come to the UK and settle down. But I knew my path was different and clear. My dream wasn’t just to be here. It was to <strong>grow, to learn, and to build a career in UI/UX</strong>, <strong>step by step</strong>. I only wanted real UK work experience in my field, not something else.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f4IpUMur9nDxBusZYdGkbA.jpeg" /><figcaption>Photo by <a href="https://www.pexels.com/photo/brown-wooden-scrabble-tiles-on-white-surface-10628657/">Brett Jordan</a></figcaption></figure><p>To get by, I took on several part-time jobs, working as a steward at football matches and events, a wine waiter, a plate waiter, a office cleaner (even while writing my dissertation!), and once, a sandwich maker at Subway. These jobs helped me survive, but I never planned to settle there.</p><p>Later, I graduated with an <em>MSc in User Experience and Interaction Design with </em><strong><em>Distinction (First class)</em></strong>, and even before graduation (right after submitting my final thesis), I had the opportunity to return to the university, not as a student this time, but as a <strong>Clinical Demonstrator</strong>. As a result, I supported the Human Computer Interaction (<strong>HCI</strong>) and <strong>UI/UX modules</strong> at Glasgow Caledonian University (<strong>GCU</strong>), sharing the knowledge I had gained while continuing to grow alongside the students.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1017/1*_8xAbu_-NfZG7QrnJxiyag.png" /><figcaption>Graduated with an MSc in User Experience and Interaction Design with Distinction in 2023.</figcaption></figure><blockquote>There’s a story about how I got my first job after my master’s as a clinical demonstrator at Glasgow Caledonian University, I’ll share that in a separate blog. But this story, this journey, is something different.</blockquote><h4>Struggling to Get Noticed in the Industry…</h4><p>While working as a clinical demonstrator at university, I continued doing part-time jobs to cover my expenses. But every day, I kept applying for jobs and faced rejection again and again.</p><p><strong>Those moments were hard</strong>. There were times when I broke down quietly, questioning everything. Because I didn’t understand why the rejections kept coming. I questioned myself several times,<em>“What’s wrong with me? Why am I not good enough?”</em> Then, I had my CV reviewed by experts and improved it little by little, but the answers never came.</p><p>Sometimes, I thought maybe it was because my visa was due to expire just three months after finishing my master’s. Maybe that was the reason no one was taking a chance on me. As I tried not to let it bring me down. It wasn’t an excuse, just a quiet thought I held onto, to give myself a reason that didn’t make me feel like I wasn’t enough.</p><p>But I held on to so much trust in myself. <strong>I knew who I was</strong>. Every time a rejection arrived, I told myself, <em>“You’re losing a great asset to your company.”</em> Maybe it sounds like confidence bordering on ego, but I needed that belief to keep going. Because <strong>I knew I wasn’t so bad that I didn’t deserve a chance</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oFbkTbZcg67C27P5ZS7s3A.jpeg" /><figcaption>Photo by <a href="https://www.pexels.com/photo/wood-light-handwritten-nature-6806399/">RDNE Stock project</a></figcaption></figure><h4>I Was Noticed by MLT Digital…</h4><p>It was January 23, 2024, when I saw a message pop up on LinkedIn from the Managing Director at MLT. He said they had emailed me the week before about a project, but it might have landed in my junk folder. Then, I checked, but the email wasn’t there. Still, he told me they had plenty of projects and asked if I could send a quote for freelance work.</p><p>That message felt like a small light appeared in a dark room. After months of searching, applying, and waiting, it was the first time in a long while that someone had reached out to me, not the other way around. And in that moment, hope returned, gently but surely.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pYnQOErWcwrqAoPlspCscA.jpeg" /><figcaption>Photo by <a href="https://www.pexels.com/photo/person-in-black-shirt-standing-on-white-floor-8347500/">Mo Eid</a></figcaption></figure><p>A few weeks later, in February, the Project Director reached out with a brief. Although it was a small project, just seven hours of work, I gave it my full attention and delivered it on time. It might have looked minor on paper, but the impact was far greater.</p><blockquote>Sometimes, it’s the smaller tasks that quietly open the biggest doors.</blockquote><p>Also, most people advised me to look for a job at a registered company, one that could sponsor my visa. I knew that MLT wasn’t registered, but I stayed in touch because what I needed was UK work experience, no matter how many years I had behind me already.</p><blockquote>A friend once told me, <em>“This country (UK) tests your patience! You don’t need all the yeses. You just need ONE.</em></blockquote><blockquote><em>One Job, One Chance.”</em></blockquote><p>That one, yes, for me, came from MLT Digital.</p><h4>Finally, I Got That ‘One’, Yes…</h4><p>In March 2024, after finishing that project, I reached out to the Project Director to say I was available for more freelance work and happy to help. Soon after, they replied with some code issues they were struggling with, so without even quoting a price or expecting anything, I offered advice on how to fix them during an online chat (Fixing code bugs is something I genuinely enjoy, something I used to do often in my previous job).</p><p>Surprisingly, they invited me to their office for a chat to understand my situation better. I accepted, not realizing it was an interview. I thought it was just a friendly chat to get to know them and maybe discuss more freelance projects while I kept searching for a full-time role.</p><p>Sitting in the cafe area, I soon saw two people I recognized from LinkedIn walk over, smiling warmly, and with a cute dog named Rizzo. Also, I’m a dog person, so that little moment instantly lifted my spirits.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_wrfrDz9LEhiGa_4zIJMbQ.png" /><figcaption>Rizzo ❤</figcaption></figure><blockquote>Indeed, the conversation felt so natural and easy. We talked about my past experience and current situation. It didn’t feel like an interview at all, but later I realised it really was. I felt completely comfortable with them, and that’s one of the reasons I love MLT’s culture.</blockquote><p>Furthermore, I explained clearly that I was teaching university modules and actively looking for a full-time role to gain UK industry experience. <strong>They were incredibly flexible</strong>, saying if I joined, I could work around my schedule. That kind of flexibility reminded me of my previous company, and I knew how much it meant for employees.</p><p>And then, in April 2024, <strong>I finally got that one, yes, I’d been waiting for.</strong> They offered me a 2 months full-time freelance role as a UI/UX designer and front-end developer (onsite) with potential extensions and the salary I asked for.</p><blockquote>Finally, that felt like a huge win. Someone trusted me, my skills, and my potential. I simply said I needed a full-time job, and they gave me one.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ksVYggIENCEi9k573Wsheg.jpeg" /><figcaption>Photo by Kindel Media: <a href="https://www.pexels.com/photo/caucasian-and-african-american-people-handshaking-7979601/">https://www.pexels.com/photo/caucasian-and-african-american-people-handshaking-7979601/</a></figcaption></figure><blockquote>After two months, I was offered a permanent position as a full-time UI/UX designer and front-end developer in June 2024.</blockquote><h3>2. From Day One to Year One</h3><p>A year ago, I walked through the doors of MLT Digital for the very first time, feeling a mix of excitement, hope, and just a little bit of nervousness. Before I even saw my manager, I was greeted by Rizzo, the office dog, tail wagging and full of charm. That moment, small as it was, put me instantly at ease.</p><p>Anyway, joining a small team meant stepping into many roles, and honestly, that was what drew me in. As a UX designer and front-end developer, I wasn’t just here to follow instructions, I was here to shape ideas, guide them from sketch to screen, and be part of something real.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6oQwuEcWiELu9YMk1LGydA.jpeg" /><figcaption>Photo by <a href="https://www.pexels.com/photo/pencil-near-white-printer-paper-273230/">Pixabay</a></figcaption></figure><p>Furthermore, what began as just another step in my career soon turned into a journey of growth, full of learning, creativity, and collaboration. Somewhere along the way, this place that once felt new started to feel like home.</p><h4>Project Highlights and Proud Moments</h4><p>Over the past 12 months, I’ve had the chance to work on real projects (Client websites, Internal websites, User analysis ect.), not just designs that look good, but ones that actually solve problems. I used a UCD approach to build websites for clients, studied how people interacted with them, improved the experience through testing, and even responded to support requests when things didn’t go as expected. Most importantly, I collaborated closely with back-end developers and directors, learning something new from each conversation.</p><p>However, the first few weeks were a chaos with login credentials, support tickets, client research, and trying to figure out how the coffee machine worked (very important, of course!). In the middle of the chaos, there was a rhythm. <strong>A team that truly cared about what they were building, and more importantly, who they were building it for</strong>.</p><p>After all, I still remember the first project I worked on. Not because it was the biggest, but because it was the moment I understood what collaboration at MLT really looked like. It was a page UX improvement task, something simple on the surface, but layered with potential.</p><p>During our morning scrum, my lead gave me a heads-up, followed by an email with the details. From day one, they trusted me to take it forward.</p><p>And that trust made all the difference.</p><p>I started with competitor research, proposed potential solutions, created wireframes and prototypes in Figma, tested them with users, and sent them for client approval. Then, the stakeholder gave the green light to a prototype with a small UI tweak that noticeably improved the user journey (they might seem like little things, but those wins stayed with me), and then I developed the page in WordPress.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QcQ7Z2xrZnC22PvflvhVcQ.png" /><figcaption>One of the Heuristic Evaluation of the client Home Page</figcaption></figure><p><strong>And the best part?</strong> Seeing the changes go live and knowing they made a difference, a feeling I’ll never forget. But that’s a story for the next part.</p><h3>3. Key Highlights from the Year</h3><p>In December 2024, as the year came to a close, we had a planning meeting paired with performance reviews. For that, I was asked to prepare a slide summarising the projects I’d worked on. Rather than just listing tasks, I saw this as a moment to share my story, through numbers, insights, and intention. A quiet way to say, “This is who I am, and this is what I’ve been trying to do.”</p><h4><strong>A UX Case Study That Quadrupled Conversions</strong></h4><blockquote>I can confidently and proudly say that I made a positive impact on the company within one year.</blockquote><p>Among all the projects, I picked two that had stayed with me, both focused on UX improvements that genuinely helped users. For that, I analysed heatmaps using Microsoft Clarity, measured click-through rates before and after updates, and pulled together the kind of insights that told a clear story.</p><p>Anyway, one project didn’t have any pre-update data, but the other did, which became my little case study. I wrote a blog about the full process and the results, which you can read on my blog.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*b54kX4bpT2Ckn8HyzYikvQ.png" /><figcaption>A 59-day heatmap analysis was conducted on the homepage (PC).</figcaption></figure><blockquote>You can read it here: <a href="https://ruwanthisulanjali.medium.com/how-ux-design-increased-a-law-firms-homepage-conversions-by-400-2f05b48649ad">How UX Design Increased a Law Firm’s Homepage Conversions</a></blockquote><p>As a result, the presentation showed a significant improvement in the client’s website, something the team noticed and appreciated. But what meant more to me was what came after.</p><blockquote>I didn’t just present the results; I offered an idea: what if we turned this into something bigger? A new service we could offer, built on data, design, and value?</blockquote><p>Furthermore, that conversation didn’t end at the meeting table.</p><p>Later, our team shared the findings with the client, and their response was overwhelmingly positive, especially around the heat map analysis. As a result, they still ask for heat map analysis regularly, which makes me quietly proud.</p><p>After all, it felt like I had contributed something meaningful, something that helped the client’s UX, and also strengthened MLT’s offering.</p><h4><strong>What Came Next: CRO Strategy Planning</strong></h4><p>Always, a small win sparked something larger.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BvVM3ehfqYxc5qda2X7feg.jpeg" /><figcaption>MLT planning meeting</figcaption></figure><p>A few weeks after our planning meeting, leadership decided to build out a CRO (Conversion Rate Optimisation) strategy that integrated UI/UX insights. First, I was invited to join strategy meetings, where I proposed a UX and performance evaluation package for clients. Then we began shaping it together, aligning our ideas to create a roadmap that could help generate more value for the business and for the clients we care about.</p><p>However, it’s still in progress, but being part of it from the start has been a milestone in itself. Once it’s finalised, our team is confident that MLT can increase revenue because of the new services , since I designed it with just a few simple steps, each of which can generate profit for the company.</p><blockquote>To think that, within one year, I not only contributed to design, but also to shaping a strategic service, that’s something I’ll carry with me.</blockquote><h3>4. Lessons Learned</h3><p>Over the past year at MLT, I’ve grown in ways that go beyond Figma files and code editors. Being a UX designer and front-end developer in a small team taught me more than just how to build or design; <strong>it taught me how to listen, adapt, and keep learning.</strong></p><h4><strong>User testing is essential</strong></h4><p>One of the most important lessons was understanding the gap between assumptions and reality. What we think users will do and what they actually do are often worlds apart. Early usability testing helped reveal those gaps. Sadly, at MLT, we don’t have a dedicated budget for formal user testing, but I made it work with internal tests, heatmap analysis, user behaviour video analysis, and feedback from teammates. Even small tests revealed big insights.</p><p>Likewise, heatmaps became one of my go-to tools. Whenever I had doubts, I’d dive into user journeys and frustrations, clicks, scrolls, hesitations. Those quiet patterns often spoke louder than words. Over time, I learned how to interpret them, fix issues, and improve the experience in meaningful ways.</p><h4>Balancing User Needs with Business Goals</h4><p>Another key learning was finding the right balance between user needs and business goals. Sometimes users need things that weren’t part of the original brief. And sometimes, businesses want quick fixes that don’t always align with what’s best for their users.</p><p>That’s where I learned the importance of communication and empathy. I started having more open conversations with clients, explaining not just what users needed, but how those changes could support their goals too. When they understood that, the conversations changed. It wasn’t about extra features, it was about value.</p><h4>Cross-functional Communication Improves Outcomes</h4><p>Most importantly, collaborating closely with developers reminded me that design doesn’t end with a prototype. In some projects, I had to wear the developer hat. Design lives and breathes in the hands of those who bring it to life. My background in front-end development became a bridge: I could speak their language, understand constraints, and adapt designs so that nothing got lost in translation.</p><p>There’s something special about knowing your design will be built exactly as you imagined it because you built that understanding together.</p><h4>Design is Never “Done”</h4><p>And finally, I’ve made peace with something I used to resist: that design is never finished. There’s always a small tweak, a hidden bug, a better way of doing something. At first, I found it frustrating, but now, I see it differently.</p><p>Some teammates jokingly call me a perfectionist. It always makes me smile, because I don’t think I am. I just like seeing things evolve. Following up on live projects, checking performance, making things smoother bit by bit, it’s become second nature.</p><p>Because in the end, design isn’t about being perfect. It’s about being present; listening, adjusting, and showing up with a better version every time.</p><h3>5. Final Thoughts: Gratitude, Growth, and the Road Ahead</h3><p>Looking back, this past year has been a quiet revolution. I’m proud of how far I’ve come, not just in the work I do with design and code, but in how I communicate, adapt, and show up with confidence I once questioned.</p><p>I’m deeply grateful to the team at MLT Digital. They didn’t just offer me a role, they offered me trust, encouragement, and a space where I feel heard. A place where curiosity is welcomed, learning is part of the process, and people genuinely care, not just about the work, but about each other.</p><p>The culture here isn’t just about ticking boxes or finishing tasks. It’s about crafting things with intention, solving problems together, and taking pride in even the smallest improvements. That kind of environment changes me. It teaches me not only how to grow, but how to belong.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*su3tPytzS_SSYjg57FsWSQ.png" /><figcaption>MLT team outing</figcaption></figure><blockquote>To anyone reading this who’s still waiting for that one ‘yes’, hold on. I know how heavy the waiting can feel. But you don’t need all the yeses.</blockquote><blockquote>Just one. One moment of recognition. One open door.</blockquote><p>And finally, to everyone who’s been part of this journey, whether you gave feedback on a portfolio, offered kind words during a rough patch, or simply believed in me, thank you. <strong>You made more of a difference than you know</strong>.</p><p>If you’re working on something meaningful, exploring a new path, or just want to chat about design, development, or navigating career changes, my inbox is always open. Let’s connect through <a href="https://www.linkedin.com/in/ruwanthi-somatilake/">LinkedIn</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f995fa06245" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[[UX Case Study] How UX Design Quadrupled a Law Firm’s Homepage Conversions]]></title>
            <link>https://ruwanthisulanjali.medium.com/how-ux-design-increased-a-law-firms-homepage-conversions-by-400-2f05b48649ad?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f05b48649ad</guid>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[data-driven]]></category>
            <category><![CDATA[user-research]]></category>
            <category><![CDATA[user-centered-design]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Mon, 24 Feb 2025 14:32:32 GMT</pubDate>
            <atom:updated>2025-07-25T20:51:47.202Z</atom:updated>
            <content:encoded><![CDATA[<h4>A data-driven approach using heatmaps, heuristic evaluation, and user testing to improve engagement and usability.</h4><p><em>— — Last updated May 2025——</em></p><p>In today’s digital-first world, a well-optimized website is essential for law firms to attract and retain clients. However, having a website alone isn’t enough; ensuring users can easily navigate, engage, and take meaningful actions is crucial for success.</p><blockquote>This is a case study I completed while working at MLT Digital (United Kingdom) as a UI/UX Designer and Front-End Developer.</blockquote><ul><li><strong>Problem</strong>: A law firm* faced a significant challenge, their homepage was underperforming, particularly on mobile devices. Analytics revealed that it was the top exit page, leading to missed opportunities for client engagement.</li><li><strong>Methods</strong>: To address this, a UX audit was conducted using <strong>heuristic evaluation</strong> and <strong>Microsoft Clarity heatmaps</strong> to identify usability issues and improve the overall user experience.</li><li><strong>Improvements</strong>: Through data-driven design improvements, including refining the <strong>CTA strategy</strong>, optimizing the homepage layout following <strong>usability goals </strong>and <strong>design principles and guidelines</strong>, and enhancing <strong>accessibility</strong>, the results were remarkable.</li><li><strong>Results</strong>: After implementing the above changes, the homepage saw a <strong>400% increase in conversions</strong>, with <strong>mobile engagement nearly tripling</strong> and significant improvements in <strong>scrolling behavior</strong> and <strong>click-through rates</strong>.</li></ul><p>This case study details the challenges faced, the research and design process, and the measurable impact of the UX enhancements.</p><blockquote>* For ethical reasons, the real name of the client or the website will not be shared. However, the process used to achieve the results will be explained, along with how the design solutions were presented using wireframes. This way, the focus is on the method and results, not the specific client.</blockquote><figure><img alt="Lawyer working at his desk in front of his laptop." src="https://cdn-images-1.medium.com/max/1024/1*xp2zlINp85f4vILphfliEw.jpeg" /><figcaption>Figure 1: Photo by <a href="https://www.pexels.com/photo/a-man-in-black-suit-jacket-using-a-laptop-8112110/">Pavel Danilyuk</a></figcaption></figure><h3>Overview</h3><ol><li>Industry Perspective on Heatmap Analysis</li><li>Client’s Requirement Analysis</li><li><strong><em>Step 1</em></strong>: Understanding the Problem</li></ol><ul><li><em>Website Heuristic Evaluation</em></li><li><em>User Research: Microsoft Clarity Analysis</em></li></ul><p>4. <strong><em>Step 2</em></strong>: Getting Client Approval and HMWs</p><p>5. <strong><em>Step 3</em></strong>: Iterative Design and User Testing</p><p>6. <strong>Results: </strong>Quadrupling Conversions</p><p>7. Client Feedback and Opportunities Gained</p><p>8. Key Takeaways</p><h3>#1. Industry Perspective on Heatmap Analysis</h3><p>Before discussing the process, the goal was to understand industry perspectives on heatmap analysis in problem-solving. The aim was to explore whether professionals use heatmap analysis or rely on different methods to analyze user behavior and identify real user problems.</p><p>To explore this, a 2-week survey was conducted on LinkedIn and Instagram. The results showed (Figure 2) that <strong>20.8% </strong>of respondents<strong> hadn’t used heatmaps but wanted to try them</strong>, while <strong>8.33% had never used them </strong>because they had other approaches.</p><figure><img alt="Do you use heatmap analysis to improve product design?" src="https://cdn-images-1.medium.com/max/600/1*vLhZGJoL4zrXCnVSNm8m4w.png" /><figcaption>Figure 2: Survey for understand industry perspectives on heatmap analysis</figcaption></figure><p>People who use heatmaps already understand their benefits and how to implement them. This article is mainly for those who want to understand how heatmaps can help improve user experience and page performance based on user behavior analysis.</p><h3>#2. Client’s Requirement Analysis</h3><p>The client’s primary concern was the <strong>underperformance of their homepage</strong>, particularly on mobile devices, as indicated by Analytics data.</p><p>A six-month analysis (Figure 3) showed that most users accessed the website via <strong>mobile</strong> rather than desktop, with the homepage being the <strong>top exit page</strong>.</p><figure><img alt="MS Clarity analysis: screenshot of page performance and device usage." src="https://cdn-images-1.medium.com/max/1024/1*3bkl7y1pSnPplU3zyBb5Ig.png" /><figcaption>Figure 3: 6-Month (Jan–June 2024) Clarity Analysis of the Client’s Website</figcaption></figure><p>As a <strong>UX designer</strong>, the goal was to identify the reasons behind this issue.</p><p>There are several ways to uncover usability problems, but considering the <strong>timeline</strong> and <strong>client’s budget</strong>, a combination of <strong>heatmap analysis</strong> and <strong>heuristic evaluation</strong> was used to diagnose the problem effectively.</p><blockquote>Since this is an <strong>existing client</strong>, prior experience working with them provided an understanding of their business and goals.</blockquote><blockquote>However, if this were a <strong>new client</strong>, it would be good to first conduct <strong>client research</strong>, considering their <strong>vision, mission, and goals</strong> to gain a deeper understanding before proceeding with the evaluation.</blockquote><h3>#3. Step 1: Understanding the Problem</h3><h4>A. Website Heuristic Evaluation</h4><p>After receiving the client’s requirements, <a href="https://www.nngroup.com/articles/ten-usability-heuristics/">Nielsen Norman’s <strong>10 heuristics</strong></a>, <a href="https://www.nngroup.com/articles/principles-visual-design/"><strong>visual design principles</strong></a>, and <a href="https://www.w3.org/TR/WCAG21/"><strong>accessibility guidelines </strong>(WCAG) 2.1</a> were used to evaluate the entire website and identify usability issues. A heuristic analysis was conducted to understand the problems across the site. Figure 4 shows the <strong>heuristic markup for the homepage wireframe</strong> on both PC and mobile.</p><blockquote>If <a href="https://www.nngroup.com/articles/how-to-conduct-a-heuristic-evaluation/">heuristic evaluation process</a> is unfamiliar, it is a review conducted by <strong>one or more experts</strong> using a <strong>set of rules</strong> to check if a solution follows them. They assess how well it meets the rules and where it falls short. Expert or heuristic evaluations depend on the evaluator’s knowledge and experience</blockquote><figure><img alt="Website and mobile view analysis using the 10 heuristics, with arrows and descriptions." src="https://cdn-images-1.medium.com/max/1024/1*6xYvNAg2WERoGAcYYWelrw.png" /><figcaption>Figure 4: Heuristic Evaluation of the Mobile and Desktop Home Page</figcaption></figure><h4>Summary of Common Issues Identified After Heuristic Evaluation</h4><p><strong><em>a. I</em>nconsistent CTA Styles and Readability Issues:</strong></p><ul><li>CTA styles and font types were inconsistent.</li><li>Mobile line height issues affected readability.</li></ul><p><strong><em>b.</em> Accessibility Concerns:</strong></p><ul><li>The “accessibility” text in the top navigation bar was not prominent enough.</li><li>The &quot;Contact Us&quot; form needed improvement; for example, placeholder text disappeared, <a href="https://www.nngroup.com/articles/form-design-placeholders/">straining users&#39; short-term memory</a>.</li></ul><figure><img alt="Smart Events showing percentage of sessions." src="https://cdn-images-1.medium.com/max/1024/1*akR4FLJLT0YCyJnzZHfscw.png" /><figcaption>Figure 5: 6-Month (Jan–June 2024), 428 sessions — Clarity Behevior Analysis for Contact Us Form Submission</figcaption></figure><ul><li>Figure 5 shows that only <strong>0.89% of users who visited the “Contact Us” page submitted the form</strong>. Improving accessibility and usability could help increase conversions.</li></ul><p><strong><em>#3 </em>Not Following Most Heuristics:</strong></p><ul><li>There is <strong>no CTA above the fold</strong>, which negatively impacts conversion rates.</li><li>Errors in the form are <strong>not easy to recover</strong> from, and every tab is marked in red, even when valid information is entered.</li><li>There are no clear <strong>next steps</strong> or instructions on what users should do after landing on the page.</li><li>Users cannot understand the images without reading the text, as the images do not convey meaning on their own.</li></ul><p><strong><em>#4 </em>Lack of a Prominent “Contact Us” CTA:</strong></p><ul><li><a href="https://www.nngroup.com/articles/contact-us-pages/#:~:text=The%20Contact%20Us%20link%20should,contact%20form%20or%20physical%20address.">Experts suggest</a> placing the <strong>“Contact Us” CTA</strong> prominently in the <strong>top-right menu</strong> or <strong>footer section</strong> for better visibility.</li></ul><p><strong><em>#5 Issues on Other Pages:</em></strong></p><ul><li><strong>Text-heavy content</strong> with minimal <strong>visuals</strong> and <strong>infographics</strong> makes information harder to digest.</li><li><strong>Lack of user trust</strong> due to missing <strong>clear navigation</strong> from service content pages to case studies and testimonials.</li></ul><p>However, this heuristic evaluation explains why the homepage became the number one exit page, as shown in Figure 3.</p><p>To address these challenges, user research was conducted. Analytics tools like <strong>Microsoft Clarity heatmaps, behavior analysis</strong>, and <strong>session recordings</strong> were used to identify where users were dropping off and what they needed to complete their tasks.</p><h4>B. User Research: Microsoft Clarity Analysis</h4><p>After conducting a heuristic evaluation, <a href="https://clarity.microsoft.com/">Microsoft Clarity heatmaps</a>, session recordings, and behavior analysis were used to identify user behavior patterns and highlight potential issues on mobile (Figure 6) and PC (Figure 7).</p><blockquote>The results were analyzed over six months data, based on 2,320 sessions, including 1,656 new users and 664 returning users across the entire website.</blockquote><figure><img alt="Colourful mobile heatmap analysis for mobile screenshots using data visualisation." src="https://cdn-images-1.medium.com/max/1024/1*ZMGR1SG_ucy5lQ_S-nHn4Q.png" /><figcaption>Figure 6: 6-month heatmap scroll and click analysis on the homepage (Mobile)</figcaption></figure><h4>[ Mobile ] 6-Month Heatmap Highlights: Scroll &amp; Click Analysis</h4><ul><li>The <strong>search bar</strong> received more clicks (Figure 6) on mobile than on PC, making it <strong>beneficial</strong> to retain for improved usability, allowing users to navigate information more effectively.</li></ul><blockquote><strong>Recommendation:</strong> <em>A </em><strong><em>clear CTA</em></strong><em> should be included in each section, including the banner; otherwise, users may lack direction.</em></blockquote><ul><li>Only one person clicked the “Contact Us” form submit button at the bottom of the page, whereas the top “Contact Us” CTA received 8 clicks (Figure 6) and directed the user to the “Contact Us” page.</li></ul><blockquote><strong>Recommendation: </strong><em>Apply the same CTA prominence to the PC view.</em></blockquote><ul><li>Only 50% of mobile users scrolled beyond the first three service boxes (Figure 6).</li></ul><blockquote><strong>Recommendation:</strong> <em>Improve the services section for mobile devices by reducing page length and enhancing visual appeal for better engagement.</em></blockquote><ul><li><strong>Session recordings</strong> show users hiding pages for extended periods, indicating distractions or loss of interest.</li></ul><blockquote><strong>Recommendation: <em>Reducing page load times </em></strong><em>(Currently 56%)</em><strong><em> </em></strong><em>and providing more engaging homepage content or clear instructions could help retain users and prevent drop-offs.</em></blockquote><figure><img alt="Colourful PC heatmap analysis for mobile screenshots using data visualisation." src="https://cdn-images-1.medium.com/max/1024/1*S__u6L79HxywI72S8GU2YQ.png" /><figcaption>Figure 7: 6-month heatmap scroll and click analysis on the homepage (PC)</figcaption></figure><h4>[ PC ] 6-Month Heatmap Highlights: Scroll &amp; Click Analysis</h4><ul><li>PC users frequently clicked on “<strong><em>OUR PEOPLE</em></strong>” and “<strong><em>CASE STUDIES</em></strong>” in the navigation bar (Figure 7), indicating a strong interest in learning <strong>about the team and success stories</strong>. Figure 8 of the <strong>Clarity analysis</strong> supports this finding, and <strong>session recordings</strong> confirm that a user navigated through various case studies, demonstrating engagement with the website’s content.</li></ul><blockquote>The behavior of new and returning users was analyzed separately. New users had more clicks on “Testimonials” and a high number of dead clicks, indicating a need to build trust and clarify information.</blockquote><blockquote>In contrast, returning users clicked more on “Our People”, ‘’Testimonials’’ and “Case Studies,” suggesting an interest in team details and success stories.</blockquote><blockquote><strong>Recommendation</strong>: <em>A clearer user journey from the homepage is recommended to improve conversions.</em></blockquote><figure><img alt="Screenshot of the top pages list taken from MS Clarity, sorted by most popular." src="https://cdn-images-1.medium.com/max/1024/1*Vahq4My9_f_j1YXpEWab9w.png" /><figcaption>Figure 8: 6-Month (Jan–June 2024) Clarity Analysis for top visted pages</figcaption></figure><ul><li>The top banner received the highest number of dead clicks despite lacking a Call to Action (CTA). <strong>Session recordings</strong> also show that some users left the website shortly after arriving on the homepage, <strong>suggesting</strong> they did not find what they were looking for or lost interest quickly.</li></ul><blockquote><strong>Recommendation</strong>: <strong><em>Implementing a primary CTA across all banners</em></strong><em>, such as “Read our Case Studies,” is </em><strong><em>advised</em></strong><em>, as research shows higher conversions for elements placed above the fold.</em></blockquote><ul><li>Scroll analysis revealed that only 15.05% of users reached the form’s submit button, with a 25% drop-off occurring after the service section.</li></ul><blockquote><strong>Recommendation</strong>: <em>As a best practice, </em><strong><em>highlighting “Contact Us” in the top navigation bar</em></strong><em> is </em><strong><em>recommended</em></strong><em> to improve accessibility.</em></blockquote><ul><li>Only 1 user clicked on the prominently displayed <strong>phone number</strong>, whereas the “Contact Us” text received 17 clicks. <a href="https://www.nngroup.com/articles/contact-us-pages/#:~:text=The%20Contact%20Us%20link%20should,contact%20form%20or%20physical%20address.">Research suggests</a> that users typically look for a “Contact Us” button rather than plain text on the nav bar.</li></ul><blockquote><strong>Recommendation</strong>: <em>On PC, replacing the “Contact Us” menu item with a button positioned to the left of the phone number could </em><strong><em>enhance usability</em></strong><em>.</em></blockquote><p>Unlike relying solely on heuristic data, real user data from the past 6 months was used to support findings and gain client approval.</p><h3>#4. Step 2: Getting Client Approval and HMWs</h3><figure><img alt="The design manager talked to the team after the affinity diagram brainstorming session." src="https://cdn-images-1.medium.com/max/1024/1*TjjYIGL4JibyEb4G5eLggg.jpeg" /><figcaption>Figure 9: Photo by <a href="https://www.pexels.com/photo/a-man-and-a-woman-in-a-meeting-3867836/">LinkedIn Sales Navigator</a></figcaption></figure><h4>Client Feedback After Reviewing the Data Analysis</h4><p>After documenting all the data, we presented it to the client at the next meeting. There were two participants from the client’s side, and they showed a lot of interest in the data we collected. They interacted and communicated actively with me and my team.</p><p>Since it was their first time seeing this kind of UIUX analysis and they were not sure what to expect (Results), they decided not to approve the full amount right away. Instead, they agreed to split the project into two phases, focusing on key improvements first, based on the available budget, and moving the rest to the next phase.</p><blockquote>As a business, we (MLT) need money, and the client needs results. If we don’t show results, we could lose the client, which means losing money.</blockquote><blockquote>I love facing challenges and this was a challenge for me to show the results clearly in the phase one and get approval for the full budget.</blockquote><h4>Finalizing the features for development</h4><ol><li><strong>How might we simplify the user journey?</strong></li></ol><ul><li>Prioritised a mobile-first design to ensure an intuitive experience for smartphone users, who make up a significant portion of the audience.</li><li>Enhanced the ‘Contact Us’ CTA in the navigation bar and footer [ <em>Moved to the </em><strong><em>second phase</em></strong><em> for consideration</em> ].</li><li>Implemented a primary CTA across all banners (above the fold).</li></ul><p><strong>2. How might we build trust and credibility?</strong></p><ul><li>Highlighted prominent client testimonials and effectively directed users to case studies.</li><li>Designed a clean, consistent, and professional interface.</li><li>Contact Us form: Ensured clear messaging and prompt, effective responses to guide users in completing tasks or understanding errors [ <em>Moved to the s</em><strong><em>econd phase</em></strong><em> for consideration</em> ].</li></ul><p><strong>3. How might we combine aesthetic and minimalist design</strong></p><ul><li>Rebranding the CTAs&#39; color — Red is associated with danger. Due to branding, red was used for CTAs, which might make users hesitate or think twice before clicking [<em>Moved to the </em><strong><em>second phase</em></strong><em> for consideration.</em>]</li></ul><p><strong>4. How might we improve scrolling and reduce dead clicks?</strong></p><ul><li>Added CTAs in each section to enhance usability and focus on conversions.</li><li>Improved the services section for mobile devices by reducing page length and enhancing visual appeal for better engagement.</li></ul><p><strong>5. How might we improve accessibility?</strong></p><ul><li>Contact Us form: Included error messages with icons instead of relying solely on colour [ <em>Moved to the </em><strong><em>second phase</em></strong><em> for consideration</em> ].</li><li>Improved colour contrast and hover effects for better visibility.</li></ul><p>After prioritizing the features, the Phase 1 design was advanced to the next stage.</p><h3>#5. Step 3: Iterative Design and User Testing</h3><p>After prioritizing the features for Phase 1, the page layout was refined using Figma.</p><blockquote>This is not a website redesign. Given the budget, the existing layout is being maintained, with a focus on UX improvements.</blockquote><p>This data-driven approach ensured that <strong>every decision was backed by real user behaviour</strong>. Here, only the wireframe design* is showcased, where both the original design and the improved version can be seen.</p><p>Additionally, items marked under ‘Phase 2’ will be considered in the next stage of development, as decided during the meeting based on the client’s budget.</p><p>In Figure 10, the <strong>Mobile homepage improvements</strong> can be seen, which were made using Figma.</p><figure><img alt="Mobile wireframe with improvements and justifications." src="https://cdn-images-1.medium.com/max/1024/1*z0JE7IwVoIUKMNcHT_E-SQ.png" /><figcaption>Figure 10: Mobile homepage improvements in Figma</figcaption></figure><p>In Figure 11, the <strong>PC homepage improvements</strong> can be seen, which were made using Figma.</p><figure><img alt="PC wireframe with improvements and justifications." src="https://cdn-images-1.medium.com/max/1024/1*KHPCrWhUguTwQSF8TS3ytg.png" /><figcaption>Figure 11: PC homepage improvements in Figma</figcaption></figure><p>After the Figma high-fidelity prototype was created and tested with two users, development began following client approval. The client approved the prototype without any further changes.</p><p>Once the development site was completed, it was tested with two users and <strong>underwent a heuristic walkthrough</strong> before being moved to the live site.</p><h3>#6. Final results</h3><p>The <strong>service section</strong>, <strong>home page banner</strong>, and overall <strong>usability and accessibility</strong> of the home page design were improved. After 59 days of the page going live, the <strong>Microsoft Clarity heatmap</strong> was analysed again to assess improvements in user behavior data.</p><h4>Mobile Conversion Results</h4><p>The following are notable results for mobile devices.</p><figure><img alt="Mobile heatmap results comparing the before and after screens with data visualisation." src="https://cdn-images-1.medium.com/max/1024/1*jcTDwyEBrZ7kS9dLUKq0og.png" /><figcaption>Figure 12: A <strong>59-day heatmap analysis</strong> was conducted on the homepage (mobile).</figcaption></figure><ul><li><strong><em>Service Section:</em></strong> Previously, the images did not effectively communicate the meaning of the title. They were replaced with icons to improve clarity, and the length of the boxes was successfully reduced. In Figure 12, you can see a <strong>significant improvement in the conversion rate</strong>, which <strong>nearly tripled</strong>.</li><li><strong><em>Top Banner: </em></strong>The top banner was identified as the most viewed section, but lacked a Call to Action (CTA). As shown in Figure 12, the mobile conversion rate is now more than <strong>four times higher</strong> than before after adding the <strong>“Read our Case Studies”</strong> CTA. This CTA was not selected at random; it was the <strong>most clicked CTA</strong> across the entire website.</li><li><strong><em>Scrolling:</em></strong> 50% of users dropped off before seeing the service section on PC. After improving the section with meaningful icons, colors, and overall page usability, there was a <strong>significant improvement in mobile scrolling behavior</strong>, as shown in Figure 12.</li></ul><h4>PC Conversion Results</h4><p>The following are notable results for PC.</p><figure><img alt="PC heatmap results comparing the before and after screens with data visualisation." src="https://cdn-images-1.medium.com/max/1024/1*KwvdOVm37oQr5nDH4rBmXw.png" /><figcaption>Figure 13: A <strong>59-day heatmap analysis</strong> was conducted on the homepage (PC).</figcaption></figure><ul><li><strong>Top Banner:</strong> As shown in <strong>Figure 13</strong>, the <strong>PC top banner conversion rate</strong> increased from <strong>0% to 6.67%</strong> after adding the <strong>“Read our Case Studies”</strong> CTA to the top banner section.</li><li><strong>Service Section:</strong> Before the improvements, the <strong>CTR in the service section</strong> was <strong>4.32%</strong>. After the enhancements, it increased to <strong>8.47%</strong>, showing a significant improvement, as shown in <strong>Figure 13</strong>.</li><li><strong>Scrolling Behavior:</strong> <strong>25% of users dropped off before seeing the service section on PC</strong>. After improving the section with meaningful icons and colors, engagement increased. As a result, a <strong>significant improvement in scrolling behaviour</strong> was observed, especially on mobile, as shown in <strong>Figure 13</strong>.</li></ul><h3>#7. Client Feedback and Opportunities Gained</h3><h4>Client Feedback After Reviewing the Results</h4><p>I first presented these results internally during MLT Digital’s product strategy meeting. Everyone, including the CEO and managing directors, was surprised by the results. I also showed how MLT Digital could earn more revenue by suggesting a CRO (Conversion Rate Optimisation) plan for other clients, which I’m now working on because everyone agreed it was a good strategy.</p><p>Later, the leadership team used this data in the client service rebrand proposal and presented it to clients in an onsite meeting. I heard from my lead that one of the senior people from the client side went back to my UI/UX data visualisation again and again, spent time reading the results, and asked questions, showing how interesting the results were for them. <strong>They were very impressed</strong>.</p><blockquote>After the meeting, our team received budget <strong>approval for the rebrand project along with phase two design improvements</strong>, which was a big win for the company.</blockquote><h4>Future works for the client and opportunities for MLT</h4><p>The next step is to work on <strong>Phase 2</strong>, keep monitoring user behaviour, and continue making improvements to the client’s website.</p><p>After the above ‘Law firm client’ UI/UX analysis, most new requests came from my lead, asking me to do similar work for other projects. They even mentioned above client&#39;s name and assigned the work directly, showing how much of a positive impact it had on the company.</p><blockquote>I’m also working on the CRO strategy plan along with the UI/UX package, which is helping MLT gain more revenue.</blockquote><blockquote>This is a big win for me as a designer, as it supports the company’s strategy and helps increase revenue.</blockquote><h3>#8. Key Takeaways</h3><p>This case study highlights the power of UX design in driving real business impact. The key lessons learned include:</p><ul><li><strong>Empathy is key</strong>: Understanding user pain points through data-driven insights leads to meaningful improvements.</li><li><strong>Data drives results</strong>: Heatmaps, analytics, and usability testing eliminate guesswork and ensure measurable success.</li><li><strong>Design impacts trust</strong>: A professional, accessible, and intuitive design builds credibility and improves engagement.</li></ul><blockquote>By prioritising UX, law firms, and any service-based business ; can turn their websites into high-performing tools that attract and convert clients.</blockquote><blockquote>Want to improve your website’s performance? Start with UX design.</blockquote><figure><img alt="Team discussion with four people about growth." src="https://cdn-images-1.medium.com/max/1024/1*lHi9BntYTlYY4rp7GspvNQ.jpeg" /><figcaption>Figure 14: Photo by <a href="https://www.pexels.com/photo/people-discussing-in-a-conference-room-7793927/">Yan Krukau</a></figcaption></figure><p>Ready to revamp your website? Let’s talk!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f05b48649ad" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[International Student Adventures: My MSc in User Experience and Interaction Design (UXID) Journey]]></title>
            <link>https://ruwanthisulanjali.medium.com/international-student-adventures-my-msc-in-user-experience-and-interaction-design-uxid-journey-7560d3660005?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/7560d3660005</guid>
            <category><![CDATA[interaction-design]]></category>
            <category><![CDATA[international-student]]></category>
            <category><![CDATA[united-kingdom]]></category>
            <category><![CDATA[masters-degree]]></category>
            <category><![CDATA[ux-design]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Wed, 03 Apr 2024 11:38:17 GMT</pubDate>
            <atom:updated>2026-03-29T17:18:22.275Z</atom:updated>
            <content:encoded><![CDATA[<h4>It was not a smooth journey, but I loved every single day</h4><p>After a one-year break from writing a blog, I have finally had the opportunity to awaken my inner writer. I graduated last November (2023) with <strong>distinction</strong> in <strong>MSc in User Experience and Interaction Design (UXID) </strong>at <strong>Glasgow Caledonian University (GCU) </strong>and decided to write this post for students who asked about my MSc journey during the design mentoring sessions on the <a href="https://adplist.org/mentors/ruwanthi-sulanjali">ADP list</a> and reached out to me through <a href="https://www.linkedin.com/in/ruwanthi-somatilake/">LinkedIn</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QnbcF-QAE9O5QBR5vF6l9A.png" /></figure><h3>Overview</h3><ol><li>How it Started</li><li>Why did I choose User Experience Masters?</li><li>Why did I choose the UK for the UXID Masters?</li><li>The MSc UXID program at Glasgow Caledonian University (GCU)</li><li>What were the difficulties I faced, and how did I overcome them?</li></ol><h3>#1. How it Started</h3><p>In 2015, I completed a <strong>Bachelor’s degree in Communication Design</strong> with a <strong>second upper class (2.1)</strong> at the <strong>University of Moratuwa, Sri Lanka</strong>. At first, I wasn’t sure about my career path, but I wanted to avoid the advertising field most of my fellow graduates were entering because I was not interested in that area. Instead, I chose the IT sector, intending to become a UI/UX designer. However, I faced challenges in UI/UX interviews because of a lack of coding skills (HTML/CSS).</p><p>Although coding skills aren’t required for UX designers, many companies in Sri Lanka expected them in 2015. So, I joined WSO2 as an intern to learn coding to enter the UI/UX industry, which was my only goal.</p><blockquote>I’m truly grateful to WSO2 because the experience I gained there helped me achieve today’s success. Additionally, my colleagues played a crucial role, and I wouldn’t be where I am today without their support. Leaving WSO2 and my colleagues was not an easy decision, but I wanted to escape from my comfort zone and achieve my goals</blockquote><p>Then, in 2018, I decided to pursue a Master’s in Human-Computer Interaction (HCI). However, financial difficulties, the program’s unavailability in Sri Lanka, and the pandemic caused a delay in this plan. Nevertheless, during this time, I engaged in deeper self-learning of UX by enrolling in online learning platforms like Udemy and the Interaction Design Foundation, due to my enthusiasm for learning UX.</p><p>Finally, in 2022, I chose Glasgow Caledonian University (GCU) to study an MSc in User Experience and Interaction Design (UXID) with four acceptance letters from other UK universities, thus bringing an end to the waiting period.</p><h3>#2. Why did I choose User Experience Masters?</h3><p>It’s no secret that UX is a growing field worldwide. After I graduated with my bachelor&#39;s in 2015, I wanted to find the best job for a designer that paid well. UI/UX was one of the top choices, matching my skills and goals perfectly. That’s why I decided to start a career in the UI/UX field.</p><p>Everyone has big life goals. To get closer to mine, I decided to do a master’s degree in UX, after getting 7 years of experience in the industry.</p><blockquote>When I was a design mentor at ADPList, one of my mentees asked, “<strong>Does he need a UX degree to work as a UX designer?</strong>” The answer is ‘<strong>No</strong>,’ you don’t have to. Everyone has a different way of thinking, even if you’re not a UX designer, and you can learn the skills and tools you need. But having a UX degree makes it easier. Based on my experience, it helps you gather lots of resources and get feedback from experts after you graduate, making it easier to get a well-paying job.</blockquote><h3>#3. Why did I choose the UK for the UXID Masters?</h3><p>Master’s programs in HCI are costly. During my master’s program research, I found interesting courses at different universities, but they were super expensive. <strong>Yes</strong>, I could have tried to get a full scholarship, and I did try, but it didn’t work, and I didn’t want to spend more time trying.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E585fAzSRTABNZf0DS8ZEw.png" /><figcaption>Glasgow Cathedral — The place that I first visited upon landing in Glasgow.</figcaption></figure><h4><strong>1. Two-Year Masters vs. One-Year Masters</strong></h4><p>I mainly thought about this when I was deciding on my master’s. A one-year Master&#39;s is the main reason for my decision to move to the UK rather than my favorite country.</p><p>Most of my favorite choices had two-year master’s programs in other countries, such as the USA, Germany, Canada, and Australia, which cost twice as much. Then I read an article by someone who had experience with both two-year and one-year master’s programs.</p><blockquote>According to them, a one-year master’s isn’t just a shorter version of the two-year program. For instance, a one-year master’s has three semesters, each with four modules, while a two-year master’s has six semesters, each with two modules.</blockquote><p>For me, doing a two-year master’s would be a waste of money and time since I had the option of a one-year master’s.</p><p>I’m not saying the UK is the only country offering a one-year UX master’s (do your research and find out what works for you and aligns with your future goals). For instance, I received an unconditional letter from an Irish university, which is a one-year program and the same cost as GCU, but I decided to go to the UK to explore the culture, that I liked the most.</p><h4>2. Reasons to choose Glasgow Caledonian University (GCU)</h4><p>I applied to five UK universities offering master’s degrees in Human-Computer Interaction (HCI) due to their diverse course content. When researching Glasgow Caledonian University (GCU), I was impressed by its vibrant learning atmosphere with cross-disciplinary practice, which aligned perfectly with my interests and international student status. When I examined the program modules, I found that many were directly applicable to the business sector, allowing me to apply theoretical knowledge to practical challenges in developing and enhancing interactive systems. These factors were key to my decision to apply to GCU.</p><p>Despite receiving unconditional acceptance letters from all five universities, I explored scholarship opportunities. While some universities offered scholarships along with their acceptance letters, I missed the deadline for applying to others. Despite GCU not being my initial top choice, its affordability with a scholarship made it an attractive option, particularly considering the course fees.</p><h3>#4. The MSc UXID program at Glasgow Caledonian University (GCU)</h3><p>I still remember my first lecture. In the morning, I had to pick up a biometric residence permit (BRB) from the post office to register for the course and get a student ID. Unfortunately, this made me a little late for the lecture, and I had to ask for help to find the right room. David McArthur (Dave) and Helen Buchanan were there, along with only a few other students.</p><blockquote>You might have heard lots about the Scottish way of speaking (It might take some time to get used to), but don’t worry, it’s not a problem during lectures. The university staff talk to students very clearly.</blockquote><p>When I got to Glasgow and went to the store, I was worried about why I had come to this country because I couldn’t understand what they were saying due to the accent. It felt like I had landed in a completely different world, and I worried about how I would understand the lectures if everyone spoke like this.</p><p>Surprisingly, during the lectures, I found that I could understand every word when Dave and Helen spoke. They were considerate and sometimes spoke slowly, making it easier for everyone to catch the information. From that point onward, both Dave and Helen have been incredibly helpful, supporting me throughout the course up until now.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*a-y1IlwRDn2Rl1zuAuqPeA.png" /><figcaption>The presentation I conducted for new MSc students at GCU for the academic year 2023-2024.</figcaption></figure><blockquote>I was also invited to deliver a webinar alongside the Programme Leader for incoming MSc UXID students, sharing my journey from student to Distinction graduate to Graduate Teaching Assistant. You can watch it here, I come in at <strong>17:58</strong>: <a href="https://www.youtube.com/watch?v=dkhUD_zjjGg&amp;list=PLTI30kaoIOjVT2EnI2MrLSsgE6jcUi44a">Watch the webinar →</a></blockquote><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FdkhUD_zjjGg%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DdkhUD_zjjGg&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FdkhUD_zjjGg%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/620e6472157d6a5fedf12ae064a25c32/href">https://medium.com/media/620e6472157d6a5fedf12ae064a25c32/href</a></iframe><p>I received numerous inquiries from individuals who asked the following questions:</p><h3><strong>#4.1 How’s the teaching staff at GCU?</strong></h3><h4><strong>1. Very supportive</strong></h4><p>During my studies, the module leaders were always there to help. They even replied on weekends, which was unexpected. One time, my supervisor called me on weekends to solve issues in my dissertation. However, you need to show your enthusiasm about your work to get their feedback. For instance, if you’re struggling with an assignment, try to do something on your own first, then meet your module leader with the questions that you have. Then they’ll guide you effectively.</p><blockquote>In design, there’s no definite right or wrong. It’s all about being creative, so there are many paths to solve one problem. Lecturers don’t often tell you what to do, but they can help clarify your direction and ensure you’re heading in the right direction. To do that, you need to make your own way and get feedback every possible time to see <strong>if you’re on the right track</strong>.</blockquote><h4><strong>2. GCU accepts student’s feedback</strong></h4><p>Every GCU student has the opportunity to give anonymous feedback on the modules. The important thing is that the lecturers listen and try to make it better next time. At GCU, students have two chances to give feedback on one module. In the middle of the semester, they provide their feedback with possible improvements from the student’s perspective. Then the module leader will go through student feedback and try to make things better by next week. Also, at the end of the semester, students give feedback about the whole module.</p><p>Now that I’m a staff member, I can see the changes that modules have made this year because of the feedback from our group.</p><h3>#4.2 How do I understand the program structure?</h3><p>When you go to the <a href="https://www.gcu.ac.uk/study/courses/user-experience-and-interaction-design">GCU MSc UXID course</a> page, you can see all the modules you’ll learn. Here’s what I understood:</p><h4><strong>1. </strong>Starting from the simplest concepts to more advanced subjects</h4><p>In our class, we had students from different fields. Not all were designers, but lawyers, software engineers, psychologists, etc. Therefore, we started with the basics and moved on to more advanced topics. In the first semester, we learned design theories, UX principles, and guidelines that are used in the industry, and they were taught in a way that everyone could understand. It was helpful for me to remind myself why each UX tool and theory matters.</p><p>In the second semester, the module focused more on developing our academic writing skills, which prepared us for the dissertation in the last semester. Also, we learned about emerging technologies like AR, VR, and AI in the ‘Advanced HCI’ module. As a result, I developed a concept for an AR visual guide for solo travelers and applied the ‘Advanced HCI’ and ‘Applied Usability Modules research methods for user testing in my dissertation.</p><p>If you learn the basics well in the first semester and learn how to research and evaluate in the second semester, you won’t need to rely on anyone for your dissertation in the third semester.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jH6qLLS-5Q04iKpD1jiy5g.png" /><figcaption>GCU MSc UXID program structure from a graduate perspective</figcaption></figure><h4><strong>2. Every module is connected</strong></h4><p>When all modules are linked together, you can’t ignore or focus too much on just one because they’re all connected. You have to use what you learn in one module for the others. For instance, I studied UX laws and user psychology theories in a module called ‘Applied User Psychology’. Then I applied those learnings to other modules such as ‘UX projects’, ‘HCI’, and ‘Visual Design’ during the first semester (I followed the same for all other modules). Finally, I used what I learned from all parts of my dissertation. This helped me get distinctions for all the modules.</p><h4><strong>3. Projects helpful for Portfolio</strong></h4><p>In my opinion, having a portfolio is super important if you’re a designer. After you finish the GCU MSc UXID course, you will have enough materials for your portfolio. If you want to be a UX designer, you can collect three projects for your portfolio from module assignments. On the other hand, if you’re aiming for a UX researcher role, you can collect more than three projects from the GCU MSc UXID course for your portfolio.</p><blockquote>I believe I performed well in all the modules because I didn’t just aim to pass. I always asked myself, “<strong>Could I show this project in a job interview after I graduated?</strong>” If I wasn’t confident about it, then I knew I didn’t deserve the highest grade. <strong>So I kept improving my work</strong>. That’s how I approached every project and ended up with top marks.</blockquote><h3>#4.3 What are the facilities for UX students at GCU?</h3><h4><strong>1. Collaboration</strong></h4><p>In UX, working together is really important when you start working in the industry. In this program, we had both individual projects and one big group project. Even for solo projects, I participated in user testing, gave feedback on other&#39;s work, and got feedback on my own work, which was also part of the collaboration. In the group project, I learned how to collaborate with UX researchers and take on roles like moderators and observers during testing. I also figured out how to make a report as a teamwork to show the client. It was challenging, but it’s vital for real-world work in UX.</p><h4><strong>2. UX Lab</strong></h4><p>GCU has a UX lab where you can find many useful things, like a room where you can observe user behavior, a device called Mr. Tappy for testing mobile apps, and software called ‘iMotions Lab’ where you can analyze users’ emotions. I went to the UX lab for a module called ‘Applied Usability’ and dissertation user tests, and it was a great experience.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/862/1*Yq-5rVimELQ61LjR4nvS-g.png" /><figcaption>(a) GCU UX Lab: living room view and observation room behind the mirror (b) Mr. Tappy. (c) iMotions emotion diagram of participants (group project).</figcaption></figure><p>I also went to a lab where GCU has a virtual reality driving simulator (VRDS Lab) to help my friend with their project and a 3D scanning and printing lab for one of my projects. The university has other cool labs too, like the Virtual-Reality and Simulation Lab (VRS Lab) and the Serious Games Lab, but I haven’t been to those yet.</p><h4><strong>3. Library</strong></h4><p>This was one of my favorite places because it was peaceful and helped me focus on my project work. It has lots of books and access to research papers about UX. On the other hand, the GCU library has quiet areas for studying alone and a floor for group discussions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/862/1*3cS_BSzUjAQbFgUWEXL-GA.png" /><figcaption>Books suggested in the modules are very helpful throughout the MSc journey and can be found in the GCU library.</figcaption></figure><blockquote>Some people asked me which laptop to buy. In my opinion, you don’t need to spend money for a laptop because the library and computer labs have plenty of desktop computers.</blockquote><blockquote>If you can’t come to university all the time, and if you can’t afford a laptop (you need to show your bank statement and expenses), <a href="https://www.gcu.ac.uk/currentstudents/fundingandfinance/fundingyourcourse/digitalinclusionitequipmentloanscheme">GCU provides laptops and dongles</a> that you can use for your entire time at university.</blockquote><h3>#5. What were the difficulties I faced, and how did I overcome them?</h3><h4>1. The GCU registration process was slow</h4><p>It had been really frustrating at the beginning of the journey. I waited a long time for the CAS request letter I needed, but thankfully, the GCU student association had made a Facebook group where you could talk to them and your batch mates before coming to university and find out how long it took to get your documents.</p><p>GCU was slow to respond because they had a lot of applications to deal with in 2022. I’m not sure if the situation has changed currently, but if you’re thinking about coming to GCU, don’t waste your time because the administration might work slowly. But once you’re in, I don’t have any complaints about the course content or teaching methods.</p><h4><strong>2. Homesickness</strong></h4><p>This is something that can happen to many international students, even if they come with family. Some people might think you can’t survive in any country if you come alone, but that’s not true. I came alone and had to figure out how to keep myself motivated.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*t019xDKh-OJeNVFVdLS45Q.png" /><figcaption>Places that I visited alone around Glasgow.</figcaption></figure><p>At first, it was tough being in my room and missing my loved ones back home. So, I made my own itinerary plan to explore and travel alone. It turned out to be a great decision because it boosted my motivation to explore this country, I spent less time thinking about my home country, and I was more focused on my studies. Eventually, my solo travel experience helped me find a research topic for my dissertation. So every experience matters; nothing is wasted. <strong>Most importantly, find your own way to motivate yourself.</strong></p><blockquote>Remember, everyone’s situation is different. I’m sharing my experience, but make sure to create your own plan based on what I’ve learned.</blockquote><h4><strong>3. Find accommodation</strong></h4><p>This wasn’t hard for me at first. I chose to stay in GCU student accommodation (<a href="https://www.gcu.ac.uk/aboutgcu/universityfacilities/caledoniancourt">Caledonian Court</a>) because it was cheaper than outside accommodations in 2022. It turned out to be a great choice because the Caledonian Court was right across from the university. Plus, it came with things like Wi-Fi, transportation, 24-hour security, bills, study room, etc. So I didn’t have to worry about those.</p><p>I know some students struggled to find accommodation when they started life here. I always advise people, especially if they’re international students, to make a plan before they come here. Bring enough money to cover expenses for at least <strong>three months without a job</strong>.</p><p>GCU also helped many of my friends find a place to live permanently and temporarily, and they provided good advice to avoid scams.</p><h4><strong>4. Finding a job (Part-time and after graduation)</strong></h4><p>This was challenging. Even in your own country, it’s not easy to land a good job. So, plan carefully before you go to any country as a student. Within three months, everyone I know found a part-time job, and I found mine in just two weeks by continually applying through Indeed.</p><blockquote>I also wrote a little something about what my journey to finding a full-time UI/UX role in the UK looked like, feel free to have a read: <a href="https://ruwanthisulanjali.medium.com/12-months-countless-lessons-my-first-year-at-mlt-digital-2f995fa06245">12 Months, Countless Lessons: My First Year at MLT Digital</a></blockquote><p>At first, it’s hard (that does not mean you can’t) to get a job in your career field, especially as an international student. <strong>You have to keep applying, expect rejection, and keep improving before applying again</strong>. It’s a cycle we need to keep repeating.</p><blockquote>Don’t worry too much about finding a job after graduating. Don’t ignore opportunities just because they seem challenging. Focus on getting good grades because that’s what you can control and keep trying. You’ll eventually get what you deserve. In the meantime, do some freelancing, volunteer in your field, and try to get some experience in the UK, even if you have work experience in your own country.</blockquote><p>GCU career services are very helpful for you to make your CV suitable for the UK format and improve your LinkedIn profile, among many other things. So it’s better to talk with GCU career advice once you arrive at GCU.</p><h3>Conclusion</h3><p>Studying abroad as an international student was tough, but it was a wise choice to step out of my comfort zone. I learned a lot from facing challenges head-on and making plans to overcome them. It’s important to understand the risks and obstacles and prepare yourself to solve them during the journey. I believe anyone can do it if they put in the effort.</p><p>The GCU MSc User Experience Design and Interaction Design course taught me a lot about UX, following industry standards, and using the latest technologies. While I wouldn’t say it’s the best program in the world, I don’t have any reason not to recommend it to anyone. However, do your own research before doing a master&#39;s degree, and do not do it for the sake of doing it. I always believe that if you follow your goals, you can get what you deserve.</p><p>Good luck with your master’s journey! I’m happy to hear your experiences and answer any questions you may have based on my own experiences :)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1017/1*jA9ijW8RVmOMAsNl3AJlUg.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7560d3660005" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ParDreamin’ 2021 Session Recording!]]></title>
            <link>https://ruwanthisulanjali.medium.com/im-a-speaker-at-pardreamin-2021-f69bd0ae0467?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/f69bd0ae0467</guid>
            <category><![CDATA[pardot]]></category>
            <category><![CDATA[dark-mode]]></category>
            <category><![CDATA[recording]]></category>
            <category><![CDATA[email-marketing]]></category>
            <category><![CDATA[speakers]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Wed, 20 Oct 2021 14:56:17 GMT</pubDate>
            <atom:updated>2021-11-10T00:13:33.720Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>My ParDreamin session on Dark Mode is now available for streaming.</strong></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FhqKzVC5mQYE%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DhqKzVC5mQYE&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FhqKzVC5mQYE%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/743f8f40fad6fa9455a01806a24bb0b8/href">https://medium.com/media/743f8f40fad6fa9455a01806a24bb0b8/href</a></iframe><blockquote><strong>Download slides </strong><a href="https://pardreamin.com/course-video/how-to-create-dark-mode-friendly-emails/"><strong>here</strong></a><strong>!</strong></blockquote><h3>Abstract</h3><p>With the majority of mobile users preferring dark mode, it is now essential that marketers, designers, and developers have this in mind when creating their content. Adding in the consideration of accessibility, email content must be easy on the eyes while still complying with company brand guidelines.</p><h3>Objectives</h3><p>The main objectives are to help understand why dark mode is important to us as email marketers and how we can improve our emails dark mode friendly.</p><h3>Topics covered</h3><ul><li>What is dark mode?</li><li>Why should you consider dark mode?</li><li>How can dark mode affect your branding?</li><li>How do different email clients treat dark mode?</li><li>How to improve your emails for dark mode?</li></ul><h3>Session highlights</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DEHy3khYoBgYfP-5f-vLvQ.png" /><figcaption>Few session feedbacks</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vx6ur0OdaAx7mTPqrV_QTA.png" /><figcaption>All set for the session!</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nGMJNATQekoHIyLK5a9m2Q.png" /><figcaption>ParDreamin’ 2021 promotion materials</figcaption></figure><p>Thank you for joining my session<strong> </strong>to learn how to make your emails dark mode friendly based on my research and experiments.</p><p>See you soon!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f69bd0ae0467" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Let’s talk about dark mode mailers.]]></title>
            <link>https://ruwanthisulanjali.medium.com/lets-talk-about-dark-mode-mailers-f2859e4a50?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/f2859e4a50</guid>
            <category><![CDATA[pardot]]></category>
            <category><![CDATA[dark-mode]]></category>
            <category><![CDATA[email-marketing]]></category>
            <category><![CDATA[speakers]]></category>
            <category><![CDATA[presentations]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Fri, 26 Feb 2021 11:36:04 GMT</pubDate>
            <atom:updated>2025-02-26T08:32:13.704Z</atom:updated>
            <content:encoded><![CDATA[<h3>[Meetup Slids] Let’s talk about dark-mode mailers</h3><h4>I’m presenting at <a href="https://trailblazercommunitygroups.com/events/details/salesforce-salesforce-marketer-group-pardot-twin-cities-united-states-presents-how-to-create-dark-mode-friendly-emails/">Twin Cities Pardot User Group</a> #TrailblazerCommunity</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MY5MbXiPFjL81LoU7czfMQ.png" /></figure><p>Please check out my recent article about <a href="https://ruwanthisulanjali.medium.com/how-i-created-dark-mode-friendly-mailers-d40e15e2a5f5">Dark mode mailers</a>.</p><p>Thank you for joining my session<strong> </strong>to learn how to make your emails dark mode friendly based on my research and experiments.</p><blockquote><strong>View slides </strong><a href="https://docs.google.com/presentation/d/1ayMpXV5kESTs4erC4EG_TKHHlAs1uYVMnRUnVlWFZMc/edit?usp=sharing"><strong>here</strong></a>.</blockquote><h4>Objectives</h4><p>The main objectives are to help understand why dark mode is important to us as email marketers and how we can improve our emails dark mode friendly.</p><h4>Topics covered</h4><ul><li>What is dark-mode?</li><li>Why should you consider dark-mode?</li><li>How can dark-mode affect your branding?</li><li>How do different email clients treat dark-mode?</li><li>How to improve your emails for dark mode?</li></ul><h4>Session images</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M7fjV2ZlGPX8od9rwoavaQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kLh4mm3caeG_GM8iDVRzyA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Dh3F5qL-AGlHnKkG_b2uzw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f2859e4a50" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Created Dark Mode Friendly Mailers]]></title>
            <link>https://ruwanthisulanjali.medium.com/how-i-created-dark-mode-friendly-mailers-d40e15e2a5f5?source=rss-6ccfc057c5e0------2</link>
            <guid isPermaLink="false">https://medium.com/p/d40e15e2a5f5</guid>
            <category><![CDATA[research]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[dark-mode]]></category>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[email-marketing]]></category>
            <dc:creator><![CDATA[Ruwanthi (Ru) Somatilaka]]></dc:creator>
            <pubDate>Sat, 07 Nov 2020 02:12:06 GMT</pubDate>
            <atom:updated>2020-11-09T03:15:21.861Z</atom:updated>
            <content:encoded><![CDATA[<h3>How I Create Dark Mode Friendly Mailers</h3><p>Dark mode is a new trend in the tech industry, popularized by its benefits including lower energy consumption and better readability. Now, most email clients support this. Many people assume that this is easy rendering.</p><p>However, it’s only true if you obtain the right route and understand the behavior of each and every email client. Let’s learn how we can make our mailers dark mode friendly from my research findings and experiments.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0prjo7Gfd8Gakzstl_xDRQ.jpeg" /><figcaption>Photo by <a href="https://www.pexels.com/@pixabay?utm_content=attributionCopyText&amp;utm_medium=referral&amp;utm_source=pexels">Pixabay</a> from <a href="https://www.pexels.com/photo/adult-anger-art-black-background-356147/?utm_content=attributionCopyText&amp;utm_medium=referral&amp;utm_source=pexels">Pexels</a></figcaption></figure><h3>What is the dark mode?</h3><p>With the new launch of iOS 13, Apple introduces the dark mode. Most popular operating systems, apps, and email clients also support dark mode. Areas with a light background are identified and converted into a dark background, and dark text becomes light.</p><p><strong><em>Do all email clients act in the same way?</em></strong></p><p>Unfortunately, some don’t respond like that. For example, the Gmail app (iOS13) client, in addition to identifying the area with the light background and turning it into the dark background, do the opposite as well (dark into light).</p><p>Before we go through the improvements let’s figure out what email clients support the dark mode as not everyone supports it.</p><blockquote>“More than 55% of emails might be opened with dark mode. Once Gmail follows the dark theme, emails that might be opened with dark mode enabled will increase to 83%”</blockquote><p><strong>What email clients support dark mode?</strong></p><ul><li>Gmail (Android, iOS)</li><li>Outlook (Android, iOS)</li><li>Outlook 2019 (Mac, Windows)</li><li>Outlook.com</li><li>Hey.com</li><li>Orange Webmail</li><li>Apple mail</li></ul><h3>Why should I consider the dark mode?</h3><h4><strong>#1 People love dark mode</strong></h4><p>I’ve run a survey on Instagram/Facebook stories and LinkedIn (175 people interacted with those questions).</p><blockquote>Based on my study, I found that <strong>overall 81.9% of people loves dark mode.</strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-iaRd3aTccxyb0cb" /><figcaption>Figure 1: Users from three different environments — Overall 81.9% of people loves dark mode</figcaption></figure><p><strong>Why do people prefer the dark mode themes that much?</strong></p><ul><li>It’s easy on the eyes because it reduces screen brightness</li><li>Most of the developers work on darker interfaces</li><li>It helps to improve content readability</li><li>Dark mode helps improve battery life</li></ul><h4>#2 The majority of people make dark mode the default setting</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*CXDcoV_NG2PwEM7G" /><figcaption>Figure 2: Users from three different environments — Overall 57% of people changed their mobile and desktop appearance into dark mode</figcaption></figure><blockquote>Based on my study, I found that <strong>overall 57% of people changed their mobile and desktop appearance into dark mode.</strong></blockquote><p>So they can check their devices with low screen brightness. This likely means that those users read their mailers in dark mode.</p><p>Some email clients have no color changes. But if dark mode users open their mailers on a web browser, we should force our mailer converted to dark mode, which is the look that most dark mode users are expecting. That’s not a negative impact on the users, we increase the readability. I’ll explain later how we can do that.</p><h4><strong>#3 Mailers look messy and it affects company branding</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*NFv-O9f91d15rBmJ" /><figcaption>Figure 3: This is how it looks like if we don’t work on the dark mode rendering</figcaption></figure><p><strong><em>What do you think about emails look?</em></strong></p><p>Exactly! It looks ugly in the dark mode and as a user, I don’t want to read the content.</p><p>This is the exact same impression that people who use dark mode will have. They won’t be happy about this. That’s why some users tend to change their device settings if they really want to read the emails.</p><p><strong>Yes, you read that correctly</strong>. Some people do that. I’m one of those users who does that too if I really want to read some email content. But this doesn’t happen all the time. <strong>Most of the time I;</strong></p><ul><li>Ignore the messy emails</li><li>Unsubscribe from my email list</li><li>Stop opening future emails from them or delete mailers without opening</li><li>Mark as spam</li></ul><p>Recently I’ve researched to find out whether people change the setting the way I did.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_7qIITHdMbTApV31" /><figcaption>Figure 4: Users from three different environments — Overall 38% of users switch the device setting when reading mailers in dark mode</figcaption></figure><blockquote><strong>Overall 38% of users switch the device setting when reading mailers in dark mode </strong>(This is bad UX).</blockquote><p>Others might unsubscribe or ignore mailers without reading. As email designers/developers, we must give a better experience to dark mode users.</p><h3></h3><p>Q2, yes! I added a drop shadow to the black logo and switched the black logo into a white logo. Use transparent images/GIFs. Avoid pure white(text)and pure black(background images). Use &quot;prefers color scheme: dark&quot; media query and changes colors and logos. #DarkModeChat</p><h4><strong>#4 Email clients respond differently</strong></h4><p>If you are a digital marketer you know that email rendering is complex. Now with the dark mode, it gets even more complex. Like I mentioned earlier, email clients work differently.</p><p><strong>Here’s how each email client works:</strong></p><ul><li>No color changes: iOS mail, Apple mail</li><li>Fully inverted: Gmail iOS app, Outlook 2019 for Windows</li><li>Partially inverted: Gmail Android app, Outlook Android app, Outlook iOS app, Outlook 2019 for macOS, Outlook.com</li></ul><p>I’ve tested how our company’s corporate colors work on each version. Therefore I’m recommending that you too test your company corporate colors to get the proper understanding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*i8L4pXJMG0Xa9lKz" /><figcaption>Figure 5: How company corporate colors work on each version</figcaption></figure><p>Here you can see how dark mode rendering affects our company branding.</p><p>Sometimes we can control it (c.), but some email clients don’t let us change the colors (b.). Therefore, it’s hard to maintain company branding with email client’s behavior.</p><p>In the next section, let’s see how we can control the dark mood to give users a better experience <strong>using a hacking method </strong>(d.).</p><h3>How can we improve our emails according to the dark mode?</h3><h4><strong>1. No color changes</strong></h4><p>(iOS mail, Apple mail)</p><p>In this version, whether the app is in light mode or dark mode your email doesn’t change. Also, <strong>we can’t force them to change it</strong>. But we can happy because our emails don’t look messy.</p><p>We only have to bother about email clients as follows.</p><h4><strong>2. Fully inverted colors</strong></h4><p>(Gmail iOS app, Outlook 2019 for Windows)</p><p>Here, it’s not only identified the area with the light background and converted to the dark background and dark text becomes light. Unfortunately, these email clients do the opposite as well: converting a dark background into a light background. Also, they <strong>won’t let us change the colors and switch the logo</strong>.</p><p>But we can only do changes as follows.</p><h4><strong>Logo</strong></h4><p>You can settle the logo with an outline, drop shadow, or background color if your logo/icon stays black (black logo is not noticeable when it changes into dark mode). According to the branding, it’s not good to change the logo color completely. Therefore I prefer to add a drop shadow to the logos and icon outlines. But we properly select the solution according to the design. So I strongly recommend testing the email before wrapping it up.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dLih7IRlIU3TtA_Wd7CDBg.png" /><figcaption>Figure 6: This is how I changed our logo for the dark mode</figcaption></figure><blockquote><strong>Tip</strong> — As I mentioned earlier we can use a white background as well. But make sure to add some <strong>extra spacing</strong> (3px width space). Otherwise, mailer does not look nice in th<em>e dark mode.</em></blockquote><h4><strong>Image</strong></h4><p>Try to use transparent backgrounds for images. Some layouts are not good with transparent backgrounds. Continuously test and change what fits for both modes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*c_tI5Hgr6ocMu5zGGAO8_A.png" /><figcaption>Figure 7: This is how we changed our images for the dark mode</figcaption></figure><blockquote><strong>Tip</strong> — Always make sure to maintain the consistency. If you are using transparent background use that throughout the layout. If you want to add a background image, make sure to add the background color.</blockquote><h4><strong>Icon</strong></h4><p>If you are using icons with black outlines, make sure to fill the icon background in white (c). Otherwise, it will not appear in the dark mode. On the other hand, if you are using full-color black icons, as follows you can either add a white outline (a) or add a white background to the icon (b). I would mostly prefer (a) or (c) because it’s jarring. But in the worst-case scenario, we can use (b) as well. Make sure to test the mail before sending it out.</p><blockquote><strong>Tip</strong> — Don’t use pure white (#ffffff) as a background color because it will be too bright and a bit hard on the user’s eyes. Alternatively, you can use an off-white like #f2f2f2 instead.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MSFE_J2bUi1lSLld4a8J2w.png" /><figcaption>Figure 8: This is how we changed our icons for the dark mode</figcaption></figure><p>Email clients are still figuring out the solution for this because this is not the right way. You can also contribute to Gmail’s dark mode <a href="https://github.com/hteumeuleu/email-bugs/issues/68">email bugs</a>. I’m pretty sure they will fix this soon. Till then we have to follow these methods.</p><h4><strong>3. Partially inverted </strong>colors</h4><p>(Gmail Android app, Outlook Android app, Outlook iOS app, Outlook 2019 for macOS, Outlook.com)</p><p>This does not function as earlier versions. Here, it <strong>only detects</strong> light backgrounds and dark text and inverts it into dark backgrounds and white text. It keeps the dark background as it is. I think that’s how the dark mode concept should ideally act.</p><p>Here are the color comparisons in each version as per the experiments I have conducted.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*w-Qw2fuyRqFzscbq" /><figcaption>Figure 8: (a) <strong>No color changes</strong>, (b) <strong>Fully inverted colors</strong>, (c) <strong>Partially inverted</strong> <strong>colors</strong>— Color comparison in each version</figcaption></figure><p>Let’s focus on the “<strong>(c)</strong> Partially inverted colors” section.</p><p>As I mentioned earlier, “(c)” only detects light background colors and dark text and changes into the dark background and light text. That’s why <strong>only</strong> color changes happen in “(e)-(f)”.</p><p>When you analyze more, you can see border-color doesn’t change while background color changes in “(e)-(f)”. I’ve added borders in the same background colors and only (c) keeps that as it is.</p><p>Sometimes dark mode users might not like default changes, there it comes the <strong>custom dark mode theme</strong> where we can change the code and colors as follows. This is kind of a hacking solution but it works very well.</p><p>Before we go through the steps, here you can understand how our mailers look after I changed the mailer into a custom dark mode theme.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kkiQU_bJ87wjwEqUDIpxIw.gif" /><figcaption>Figure 9: After mailers developed into a custom dark mode theme</figcaption></figure><h3>How I improved the default dark mode style</h3><h4><strong>#1 We should enable the dark mode for email clients</strong></h4><p>First, we should add meta tags as follows into your mailer &lt;head&gt; tag.</p><pre><em>&lt;meta name=”color-scheme” content=”light dark”&gt;<br>&lt;meta name=”supported-color-schemes” content=”light dark”&gt;</em></pre><h4><strong>#2 Add “prefers-color-scheme: dark” media query</strong></h4><p>Then, you have to add “prefers-color-scheme: dark” as a media quarry into your mailer &lt;style&gt; tag.</p><p>We have a “view online” feature on our mailers. If dark mode users will open our email on the web browser, we need to give them a better experience. Therefore I’m changing the email colors and logos using the following media query.</p><pre><em>&lt;style&gt;<br>@media (prefers-color-scheme: dark) { }<br>&lt;/style&gt;</em></pre><p>But some email clients don’t support that, here you can see the <a href="https://sidemail.io/articles/dark-mode-in-html-email/">chart</a> where I get it from side mail blog. (as of October 2020)</p><p><strong>Email’s client support “prefers-color-scheme: dark”</strong></p><ul><li>Apple Mail — iPhone + iPad</li><li>Outlook — macOS</li><li>Outlook.com — webmail</li><li>Mozilla Thunderbird — Windows 10</li><li>Spark — macOS, iOS 13, Android 9</li></ul><p><strong>Email’s clients DO NOT support “prefers-color-scheme: dark”</strong></p><ul><li>Outlook — iOS 13, Android 10, Windows 10</li><li>Gmail — Android 10, iOS 13, webmail</li><li>Apple Mail — macOS</li><li>Yahoo! — webmail</li></ul><p>Now let’s see how we can add CSS styling into the media query.</p><pre><em>&lt;style&gt;</em></pre><pre><em>@media (prefers-color-scheme: dark) {</em></pre><pre><em>.</em><strong><em>dakmode</em></strong><em> {<br>background-color: #202123 !important;}</em></pre><pre><em>.</em><strong><em>darkLogo</em></strong><em> {<br>display: none !important;}</em></pre><pre><em>.</em><strong><em>lightLogo</em></strong><em> {<br>overflow: visible !important; <br>display: block !important; <br>width: auto !important; <br>max-width: inherit !important; <br>float: none !important; <br>max-height: inherit !important; <br>visibility: inherit !important; }</em></pre><pre><strong><em>P, h2, span, li</em></strong><em> {<br>color: #d2d2d2 !important;}</em></pre><pre><em>.</em><strong><em>link:hover</em></strong><em> {<br>color: #ffffff !important}</em></pre><pre><em>}</em></pre><pre><em>&lt;/style&gt;</em></pre><p>I’ve avoided pure white (#ffffff) as the text color and pure black (#000000) as the background color because I’ve found that it’s a bit harder on our eyes and it’s nice to keep colors between not too bright or not too dim.</p><blockquote><strong>Tip</strong> — However, I’ve used pure white (#ffffff) as a hover effect to the text link. It helps to grab the user’s attention to the links.</blockquote><h4><strong>#3 Add styles to the email body</strong></h4><p>Always make sure to add the dark mode class into your HTML body. Otherwise, your changes won’t be visible.</p><p>First, you should add a “.darkmode” class into the HTML &lt;body&gt; tag as follows.</p><blockquote><strong>Tip</strong> — Here, don’t include “background-color” inside the style. You have to make sure to take it out from the style. Otherwise, mailer won’t show the changes.</blockquote><pre><em>&lt;body </em><strong><em>class=”darkmode”</em></strong><em> </em></pre><pre><em>style=”font-family: ‘Roboto’, Helvetica,sans-serif; -webkit-text-size-adjust: 100%; -ms-text-size-adjust: 100%; height: 100% !important; width: 100% !important; margin: 0;padding: 0;” data-gr-c-s-loaded=”true” </em></pre><pre><strong><em>bgcolor</em></strong><em>=”#FAFAFA”&gt; </em></pre><pre><em>&lt;/body&gt;</em></pre><p>Then you can work on the logos. Here you can see how “.darklogo” and “.lightlogo” classes inserted. This appears in your dark mode vs. light mode logos.</p><pre>&lt;tr&gt;<br>&lt;td&gt;</pre><pre>&lt;a href=”https://wso2.com/&quot; style=”text-decoration:none;” target=”_blank”&gt;</pre><pre>&lt;img alt=”WSO2 Logo” <strong>class=”darkLogo</strong>” src=”<a href="https://wso2.cachefly.net/wso2/sites/all/images/2020/wso2-logo-blacklogo-darkmode.png">https://wso2.cachefly.net/wso2/sites/all/images/2020/wso2-logo-blacklogo-darkmode.png</a>&quot; style=”width: 100px; -ms-interpolation-mode: bicubic;height: auto; outline: none; text-decoration: none; border: 0;”&gt;</pre><pre>&lt;/a&gt;</pre><pre>&lt;<strong>div</strong> align=”center” <strong>class=”lightLogo</strong>” <br>style=”visibility:hidden; width:0px; max-width:0px; max-height:0px; display:none; overflow:hidden; float:left; line-height:0px;”&gt;</pre><pre>&lt;img alt=”WSO2 Logo” <strong>class=”lightLogo</strong>” src=”<a href="https://wso2.cachefly.net/wso2/sites/all/images/wso2-logo-white-new.png">https://wso2.cachefly.net/wso2/sites/all/images/wso2-logo-white-new.png</a>&quot; style=”visibility: hidden; width: 0px; max-width: 0px; max-height: 0px; display: none; overflow: hidden;float: left;line-height: 0px;”&gt;</pre><pre>&lt;/<strong>div</strong>&gt;</pre><pre>&lt;/td&gt;<br>&lt;/tr&gt;</pre><h4>#4 Always test your emails</h4><p>We have to always test our mailers before sending it out. Also, you must do the AB testings because email clients are changing, and every time it affects the user experience.</p><h3></h3><p>Q4, Through research and AB testing we can show them the result. Also by analyzing click-through reports show the better engaged in our mailers. #DarkModeChat</p><h3>Conclusion</h3><p>Since the dark mode is a trending feature, most of the email clients start supporting that. But the biggest problem is that email rendering can be complex with dark mode, and email clients act differently.</p><p>On the other hand, the majority of people love dark mode and have set it as their default mode. Therefore we have to give a better experience while they are reading emails in dark mode.</p><p>As email designers/ developers, we only have to focus on the logo, images, icons, colors (text and background), and custom code (use <strong>“</strong>prefers-color-scheme: dark” media query<strong>)</strong>. If we’ve done those necessary changes, we can give a better experience to the dark mode users, and it helps to increase the open rate, CTR, CTOR, and read rate of our mailers.</p><p>All the best for your next email campaign and I would love to hear your experiences. :)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0gzGEGTygZClSfnwFgaCiA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@walling?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Walling</a> on <a href="https://unsplash.com/s/photos/dark-mode?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d40e15e2a5f5" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>