<?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 Simarjeet Singh on Medium]]></title>
        <description><![CDATA[Stories by Simarjeet Singh on Medium]]></description>
        <link>https://medium.com/@simarjeetsingh?source=rss-192e291131d9------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*_QNK5R7O-xz7nFVJODu_5Q.jpeg</url>
            <title>Stories by Simarjeet Singh on Medium</title>
            <link>https://medium.com/@simarjeetsingh?source=rss-192e291131d9------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 23:18:16 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@simarjeetsingh/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[AI Studio]]></title>
            <link>https://medium.com/@simarjeetsingh/ai-studio-cbc82305edd6?source=rss-192e291131d9------2</link>
            <guid isPermaLink="false">https://medium.com/p/cbc82305edd6</guid>
            <category><![CDATA[design-process]]></category>
            <category><![CDATA[bots]]></category>
            <category><![CDATA[bot-platform]]></category>
            <category><![CDATA[assistant]]></category>
            <dc:creator><![CDATA[Simarjeet Singh]]></dc:creator>
            <pubDate>Tue, 25 Jul 2017 07:07:38 GMT</pubDate>
            <atom:updated>2018-06-03T01:09:40.163Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NCauJErdbmsY2RCQ0xyRFA@2x.png" /></figure><h4>Yahoo’s platform to create intelligent conversational assistants.</h4><p>A conversational assistant needs to understand what a user is saying, what he means when he says that and what to do when he means that. Today, there are platforms that let us build assistants trained to have all these capabilities. Yahoo’s AI studio is one such platform. It was originally conceived as a public platform but is now used exclusively to build Yahoo’s own bots, skills and assistants.</p><h3>My Role</h3><p>I designed the overall navigation and information architecture for the platform and the ‘Flows’ functionality in detail among other smaller parts.</p><h3>Navigation</h3><p>When I joined the team, they had built a number of features for the platform but had not focused on it&#39;s information architecture and navigation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*itfgyO_XpLvCAe-ahatHfQ.png" /><figcaption>Actions and tasks were sprinkled all over the page and lacked a clear hierarchy when I reviewed this existing design.</figcaption></figure><p>I took a deep look at various components that made up the platform and how they worked together and also undertook a competitive analysis study reviewing other platforms like Meya, Api.ai and Wit.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eqO74fLogYgHKLTvPwnHoA@2x.png" /><figcaption>System diagram created to review my understanding of the platform with other stakeholders.</figcaption></figure><p>Combining these learnings, the new IA and navigation emerged.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1Mf8K2ePGryd7HMgM2GklQ@2x.png" /><figcaption>New Navigation</figcaption></figure><h3>Need for Flows</h3><p>We realized that while our platform made it easier to train an assistant to understand a user’s utterances and provide simple answers, it was unable to handle complex business logic or have a meaningful conversation.</p><blockquote>“What time is my meeting with Alex?”<br>(Bot) “You are meeting Alex at 9:00am today”<br>“Can you ask him to move our meeting to 10:00am instead”<br>(Bot) “<strong>Whom should I ask?</strong>”</blockquote><p>There needed to be a way to keep track of context in a conversation and convert product flowcharts into logic.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1008/1*bkJytdLDPlNnPtqHJedmNQ@2x.png" /><figcaption>Flows can easily convert such flowcharts into assistant logic.</figcaption></figure><h3>Design for Flows</h3><p>I transferred my initial understanding of what flows were into the first design iteration. Feedback from users <em>(developers, trainers and product managers)</em> was that while it was easy to understand, it did not make context management much easier.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qVy4Q89PF-o6sgxUinTOrA.gif" /><figcaption>First Iteration</figcaption></figure><p>The second iteration broke flow creation and flow visualization into two separate views. It also emphasized that a flow was made up of triggers and states. Triggers made an assistant to go from one state to another. This concept got traction with users and I started refining it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UwALVwb7sdpvycdj-SX3wA.gif" /><figcaption>Second Iteration</figcaption></figure><h3>Refinement</h3><p>A few cycles of proposals, feedback and refinement were spent to figure out the right way to create a flow and the best way to visualize it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NaZgdFyn1rWOoIg5ierhKA.gif" /><figcaption>An exploration on how to shift between view mode and edit mode.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2UMxtQign8c9f9uVitoHgQ.png" /><figcaption>Flow visualization concept, rejected for showing too much complexity upfront.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*otfN-tTqBftoYObRScmPYA.png" /><figcaption>Flow visualization concept, rejected for looking complex and not being scalable.</figcaption></figure><p>Finally a combination was achieved that seemed to perform well on all considered parameters.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6oqcuETeelqgn0AX0byrMA@2x.png" /><figcaption>Finalized flow visualization</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Rg2EyBkfUBcbuDTWV_71zQ.png" /><figcaption>Finalized flow editing form</figcaption></figure><h3>Looking Ahead</h3><p>Releasing AI Studio to the public <em>(as was originally planned)</em> would have given us a lot of insights on how people were using it and where it needed improvements. Although number of users are limited inside Yahoo, we plan to keep learning from them and keep improving the platform further.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cbc82305edd6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[5 still relevant learnings]]></title>
            <link>https://medium.com/@simarjeetsingh/5-still-relevant-learnings-cfead5c3c8e3?source=rss-192e291131d9------2</link>
            <guid isPermaLink="false">https://medium.com/p/cfead5c3c8e3</guid>
            <category><![CDATA[design-process]]></category>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[designs-gadget]]></category>
            <dc:creator><![CDATA[Simarjeet Singh]]></dc:creator>
            <pubDate>Fri, 21 Jul 2017 00:38:34 GMT</pubDate>
            <atom:updated>2017-09-29T19:28:17.332Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wyNj5Kt11ZrMAxih4hOCBA.png" /></figure><h4>From designing a bunch of gadgets almost a decade ago.</h4><p>I designed a health monitor, a thermostat and a home energy manager during my days as a designer at Honeywell. While these products themselves have evolved or become obsolete over the years, what I learnt in those days has stayed with me and guided my work since. Let me share those learnings with you.</p><h3>The Gadgets</h3><p>A health monitor that recovering patients used to measure their vitals and share those with nurses tracking their progress. A popular home thermostat adapted for commercial and industrial environments. An interactive device to manage household energy consumption.</p><h3>1. Model the problem</h3><p>The first time that I tried to model a design problem was when designing the home energy manager. We had narrowed down a list of features for the product.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/803/1*ptK-IfpCPII6TjBrn6900w@2x.png" /></figure><p>A day went by and I <strong>struggled</strong> to convert these features into a user interface. At this point, I drew a simple diagram of what information would a user see on this device and how could she react to it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/776/1*7nJdqpzKMkrtlrRHgM0oxg@2x.png" /></figure><p>This allowed me to have a framework to categorize product features and subsequently translate them into a user interface. I have used models in some way or the other in every significant project ever since.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/776/1*gf-M24wOXbg1XuFLtSnQBA@2x.png" /></figure><h3>2. Use Analogies</h3><p>Unlike homes, multiple thermostats need to be installed at an industrial site. These thermostats need to communicate with each other to transfer settings and user actions. They also need to allow themselves to be remotely accessed. To design how this communication could materialize, I explored how motifs from analogous systems could be adapted to this system.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YBALIJLLsoWTTPzsAPzlLQ@2x.png" /></figure><p>This exercise gave me a large number of approaches to evaluate. The final proposal was developed after studying the pros and cons of all these approaches and combining the best parts of the better fits.</p><h3>3. Prototype</h3><p>Whether to communicate the initial concept for home energy manager or to detail out a feature for the industrial thermostat or to field test a proposal for health monitor with end users, prototypes have been my preferred design communication partners. They have proved more effective than flow diagrams and detailed mocks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SybVPStp7CF52OHAUlr41g.gif" /><figcaption>Initial concept for home energy manager</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ls8Z4NqWy0YDvuq9cNAUJw.gif" /><figcaption>Prototype showing set point and schedule change flows on thermostat.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/667/1*u_d4AprsFcJ136qwOe2iKw.gif" /><figcaption>Interactive prototype used for gathering user feedback on health monitor.</figcaption></figure><h3>4. Listen to ALL the users</h3><p>Health monitor users were required to measure their body vitals at fixed times during the day. These numbers would be send to the nurses and health care providers to track their progress. I talked to a number of end users of health monitor and included their feedback into the product. One of their suggestions was to introduce the ability to pause a vital measurement session.</p><blockquote>“It is really inconvenient when someone shows up at the door and I can’t open it immediately as my session is in progress.”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vTycFpr-oJDPlw6K1nZLIg.png" /><figcaption>The contentious pause button</figcaption></figure><p>However as I showed the modified prototype to nurses at a later meeting, they wanted the pause button to be removed.</p><blockquote>“We don’t want patients to feel that there is something else more important than their health.”</blockquote><p>It was a good lesson for me to check for competing requirements before committing myself to a design change.</p><h3>5. Question the findings</h3><p>We had a premise for a home energy manager and conducted an elaborate field study that supported it. We then created a product and found out that <strong>people were not willing to pay for it</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hV3mwAYSPs-BlP2ygn5EqQ.png" /><figcaption>A field study review meeting in progress</figcaption></figure><blockquote>Did people have problems but also simpler ways of handling them?<br>Did people not perceive enough ROI from our product?<br>Did we ask users the right questions?<br>Was this even a management and monitoring problem?</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*JpWRyOmV4PctDRCODthq8w.png" /><figcaption>Home energy manager (2008)</figcaption></figure><p>We should have spent more time questioning our assumptions and defining the problem.</p><h3>Conclusion</h3><p>A lot has changed since I designed some gadgets at Honeywell. Mobile phones are now ubiquitous, mobile apps have redefined interface design norms, we are at the cusp of an A.I. revolution, conversational assistants and voice interfaces are slowly but surely taking over as the de facto way of communicating with technology… and in all this I find learnings from my formative years as relevant today as they were 8 years ago.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cfead5c3c8e3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Designing Search Experiences]]></title>
            <link>https://medium.com/@simarjeetsingh/designing-search-experiences-fc75d504ce5d?source=rss-192e291131d9------2</link>
            <guid isPermaLink="false">https://medium.com/p/fc75d504ce5d</guid>
            <category><![CDATA[search]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[search-design]]></category>
            <category><![CDATA[design-process]]></category>
            <category><![CDATA[information-design]]></category>
            <dc:creator><![CDATA[Simarjeet Singh]]></dc:creator>
            <pubDate>Mon, 03 Jul 2017 23:34:15 GMT</pubDate>
            <atom:updated>2017-08-27T20:55:45.452Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QRh_U10D9Y9JPVogmlRotg@2x.png" /></figure><h4>Finding what really matters to users helps inform design decisions.</h4><p>Every day, millions of people make billions of search queries for thousands of entities like hotels, restaurants, stores, places to visit, products, health issues, movies, autos and programming questions.</p><p>In addition to providing algorithmic results <em>(blue links)</em> on each search results page, today’s search engines provide custom search experiences for most entities. These custom experiences should aim to reduce cognitive load on users and save their time by collating the most relevant information and presenting it in a clear and concise manner.</p><h3>What matters</h3><p>Each entity has a unique set of features that define it. All these features are not equally important and can be classified into 3 kinds based on their relevance to users:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e47MSV7wMc6eahPMOMzD0g.png" /><figcaption>3 kinds of features</figcaption></figure><p>While designing search experiences, essential features should be highlighted. More real estate should be given to features that augment the essentials and insignificant features should be ignored. Below are examples from my work on Yahoo search where I have used this simple design approach.</p><h3>Hotels</h3><p>I conducted a user study to determine that given a list of hotels, what factors helped users to decide which hotel to book. While it was clear that cost, star rating, hotel pictures and amenities were essential or augmenting factors, users did not find hotel address to be useful information.</p><p>Most users search for hotels in cities they don’t live in. They can’t figure out if 72 Rue De Turbigo, Paris is a good location or not.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VcWlO-F7Zw0Yi3MEVZgHyQ.png" /><figcaption>Both options above represent the address of the same hotel. Which information do you find more actionable?</figcaption></figure><p>Giving a context of how far a hotel is from the nearest Point of interest, Airport or Subway station helps users picture if that hotel suits them or not.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jJ6vW6bZBX5qOY6O-gHmgg@2x.png" /><figcaption>Actionable hotel location information in this hotel listings design for mobile.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wabYz0Of3A0bKFY2I42BUQ@2x.png" /><figcaption>Maps transformed into an active decision aiding feature instead of being just passive information.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fw2cJsoUjpyjgLxfVRW5GA@2x.png" /><figcaption>Location module at the bottom of the hotel details section helps put the hotel’s location in context.</figcaption></figure><h3>Programming</h3><p>I enlisted a group of my programmer colleagues to list down their recent programming queries <em>(limited to HTML, CSS and Javascript)</em>, share the intent behind those queries and the resources from where they got their answers.</p><p>All their queries could be classified into two buckets, ‘reference’ and ‘community reach out’. <a href="https://developer.mozilla.org/en-US/">MDN</a> and <a href="https://stackoverflow.com/">Stack Overflow</a> were the preferred resources for these two buckets.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/968/1*zrii6-yZAaD9PDjtQyb1LA@2x.png" /><figcaption>Classification of various query intents into two buckets</figcaption></figure><p>Next, I collected and added up their inputs on what information in these resources was useful to them in finding their answers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/968/1*hRwkwSl_Q4qKHZXf1BLblA@2x.png" /><figcaption>A questionnaire to capture what matters to users for community reach out queries.</figcaption></figure><p>A visualization of what mattered to most people most of the time emerged.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/454/1*ZF86F5iiv1ssQo2xRn3IWQ@2x.png" /><figcaption>Relevant information for most queries in the two buckets</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/659/1*rCCeb1O2i1wtgATYKLkA-g@2x.png" /><figcaption>Designs for reference queries in CSS, HTML &amp; JS.</figcaption></figure><p>Initial research is not always conclusive but it does help to narrow down to a few options that can then be tested in the field.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*cDvMuIFwzMpCY_jBtNZlFg@2x.png" /><figcaption>Concepts for community out reach queries created using learnings from user study. Further experimentation was needed to determine which approach was more suitable.</figcaption></figure><h3>Health Conditions</h3><p>We found that a significant number of queries made on our search engine were for health conditions like migraine and flu. Using opaque query logs with no user information, we analyzed the most popular queries that people used to search for health conditions and tried to elicit the intents behind those queries.</p><p>We found that most people were looking to check or confirm symptoms. Some people were also looking for self medication and tended to land on sites with questionable credentials.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_kDJzjN8ffKmekg4WGQOrA@2x.png" /><figcaption>The resultant design reflects the author’s understanding of user needs.</figcaption></figure><h3>Local businesses</h3><p>When searching for local businesses, people found the combination of Location and Timings as most useful information.</p><p>Consider that you want to go to a furniture store 7 miles away and that it closes in 55 minutes. Would you still go there? To answer this question, you would do a mental calculation <em>… 20–25 minutes to reach, 5–10 minutes to park, I am left with 20–25 minutes. Does this much time work for me?</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EKrGYyZHPmfuAj7D-n8a4A@2x.png" /><figcaption>Prominent distance and timings information in these listing designs</figcaption></figure><p>When searching for chain businesses near them, users already know which business they are searching for and there is no point in repeating the same business name in each listing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OwIbwnNFpX48ELv8qr3ydA@2x.png" /><figcaption>Distance and department information is more useful than names for chain businesses.</figcaption></figure><h3>Destinations</h3><p>I analyzed the search results that (anonymous) users clicked on the search results page after they searched for points of interest or cities and found that majority of users clicked on results for Travel information websites and Hotel and flight booking websites. This suggested a strong travel intent.</p><p>A similar analysis for country queries showed that majority of users clicked on wikipedia and other informational websites suggesting a strong exploration intent.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4v90CS6pX_c0w3clpCE3AA@2x.png" /><figcaption>Selection of information nuggets and links to explore more made based on majority user intents in these Knowledge graphs.</figcaption></figure><h3>Conclusion</h3><p>The design approach illustrated above tries to make a good guess of what might work for users. Real fun begins when we put these designs out, observe how users react and strive to continuously learn and improve.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fc75d504ce5d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Highlight the relevant, ignore the others.]]></title>
            <link>https://medium.com/@simarjeetsingh/highlight-the-relevant-ignore-the-others-b7ad96a4d9f1?source=rss-192e291131d9------2</link>
            <guid isPermaLink="false">https://medium.com/p/b7ad96a4d9f1</guid>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[design-process]]></category>
            <category><![CDATA[lists]]></category>
            <category><![CDATA[search]]></category>
            <category><![CDATA[design]]></category>
            <dc:creator><![CDATA[Simarjeet Singh]]></dc:creator>
            <pubDate>Sat, 15 Apr 2017 01:55:42 GMT</pubDate>
            <atom:updated>2017-06-21T21:30:35.004Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w2CV8MXGMjoYQsYPtePIdg@2x.png" /></figure><h4>A strategy to design destination search listings.</h4><p>Every day, millions of people search for physical destinations with queries like ‘restaurants near me’, ‘hotels in Paris’, ‘furniture stores in Sunnyvale’ or ‘McDonalds near me’. Their search experience for all such queries starts with a list of recommendations. The factors that help people decide which amongst the listings to explore further and which to ignore vary based on the category of destinations they are searching for. Knowing these factors helps us design more scannable lists. Easier scanning leads to users exploring more listings which ultimately leads to more bookings and reservations.</p><h3>A shared vocabulary</h3><p>Physical destinations share a common vocabulary of features that help define them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/968/1*chO4VKnb4eW6nFFdVifpiQ@2x.png" /></figure><h3>What matters</h3><p>While the feature vocabulary is shared, features that really influence a user’s decision to explore more about a destination or not vary based on destination category. Features can be classified into 3 kinds for each category, based on their criticality to users.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*p_Lj5fh_HIw1pxSW9JLkCw.png" /><figcaption>3 kinds of features</figcaption></figure><p>While designing search listings, essential features should be highlighted. More real estate should be given to features that augment the essentials and insignificant features should be ignored.</p><h3>Hotels</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/968/1*mDDvzJqejPUFjEquYYgO2A@2x.png" /><figcaption>Relevant factors for deciding on a hotel</figcaption></figure><h4><strong>Cost</strong></h4><p>A user with a cut-off budget of $200/night, may perceive Hotel A to be a more attractive deal than Hotel B.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nB4BUcW7iNYhvXgf8zzPPA.png" /></figure><h4><strong>Quality</strong></h4><p>People mentally calculate the quality of a hotel by considering both it’s rating and number of reviews. Users may consider Hotel A to be better than Hotel B. However both Hotel A and B should be better off than Hotel C.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EEMZU38d6IkX-Jr8TbMzgA.png" /></figure><h4><strong>Location</strong></h4><p>Most users look for hotels in cities they don’t live in and are not aware of it’s neighborhoods. They can’t figure out if 1100 Tasman drive, San Jose is a good location or not.</p><p>Hotel address itself is not actionable information. Giving a context of how far the hotel is from the nearest Point of interest, Airport or Subway station helps users picture if that hotel suits them or not.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VcWlO-F7Zw0Yi3MEVZgHyQ.png" /><figcaption>Both options above represent the address of the same hotel. Which one do you find more informational?</figcaption></figure><h4><strong>Ambience</strong></h4><p>A user may not book a hotel with good user rating, price and location if he does not like its pictures. However pretty pictures can’t make a user book a hotel poor in other factors. Detail picture viewing experience should therefore be shown only after a user expresses interest in a hotel by selecting it’s listing.</p><h4><strong>Amenities</strong></h4><p>Amenities usually come either at the beginning of the search process where users use them as a filtering criteria <em>‘show me only hotels with breakfast included’</em> or act as deal breakers at the end if a hotel is not so good in the specific amenities that matter to the user, say he reads a review that says ‘<em>oh this place has terrible breakfast</em>’.</p><h4><strong>Name</strong></h4><p>Name is essential to identify one hotel from another (we will see later how this is not true for some other destination types.)</p><h4>Resultant Design</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*u3Xfza1I5SK0l5zQ0O2I9A@2x.png" /></figure><h3>Local businesses</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/968/1*5_jAPgzLRfq55BLm8IZ7qQ@2x.png" /><figcaption>Relevant factors for deciding which local business to go to</figcaption></figure><h4><strong>Location &amp; Timings</strong></h4><p>Suppose a furniture store is 7 miles away from you and it closes in 55 minutes. Would you still go there? To answer this question, you would do a mental calculation <em>… 20–25 minutes to reach, 5–10 minutes to park, I am left with 20–25 minutes. Does this much time work for me? </em>If 20 minutes is not enough time for you, you may re-schedule your visit or explore other listings.</p><h4>Cost</h4><p>$, $$ or $$$ gives a lot of information about a restaurant’s price range. However the price range of your neighborhood food market or optometrist cannot be so easily abstracted. Price range is therefore relevant for only a few types of listings.</p><h4>Name and Quality</h4><p>Just like in Hotels, both Name and Quality are essential factors.</p><h4>Ambience and Menu</h4><p>Ambience and Menu are more important after a user has decided to explore a listing and help finalize one restaurant over another. For other types of businesses, ambience has little relevance and menu none.</p><h4>Resultant Designs</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EKrGYyZHPmfuAj7D-n8a4A@2x.png" /></figure><h3>Chain businesses</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/968/1*GD-ZbM0ADbp-q9Ay_6QPuw@2x.png" /></figure><p>Unlike hotels and local businesses, when users search for a chain business, they have already made up their mind on going there<strong>*</strong>.</p><p><em>*Many users intend to visit online sites of chain businesses instead of going there physically. This is a very different use case and is not the focus of this article.</em></p><h4>Location, Timings and Quality</h4><p>Convenience takes precedence over quality. Users are more inclined to visit the USPS office that is open and closest to them than look for a USPS office with best reviews.</p><h4>Departments</h4><p>Say a user searches for ‘Bank of America near me’. Highlighting information that the closest BofA is an ATM and the closest open BofA Financial Center is 3.8 miles away should save some time to a user trying to cash a cheque.</p><h4>Name</h4><p>Name is <strong>not useful</strong> to differentiate one listing from another.. The user already knows which business he is searching for. There is no point in repeating the same business name in each listing.</p><h4><em>Resultant design</em></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*t3Wo0WWyC_1oqEjhl423cQ@2x.png" /></figure><h3>Conclusion</h3><p>Highlighting information that people <strong>really</strong> use to choose a destination over others can help design listings that streamline a user’s destination selection process. An easier selection process leads to more bookings and more satisfied users.</p><p><em>The author has helped design Yahoo Search’s travel and local search experiences. The ideas shared in this article are collated from his experiments and research and has informed his own designs for these products.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b7ad96a4d9f1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Election day with Monkeypet]]></title>
            <link>https://medium.com/@simarjeetsingh/election-day-with-monkeypet-b6591b11128c?source=rss-192e291131d9------2</link>
            <guid isPermaLink="false">https://medium.com/p/b6591b11128c</guid>
            <category><![CDATA[dialog-design]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[flow-design]]></category>
            <category><![CDATA[bots]]></category>
            <category><![CDATA[bot-design]]></category>
            <dc:creator><![CDATA[Simarjeet Singh]]></dc:creator>
            <pubDate>Sun, 22 Jan 2017 16:38:13 GMT</pubDate>
            <atom:updated>2017-09-29T20:39:03.649Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-KNLTNZtbcS9Vc7xElvOdw.png" /></figure><h4>How Yahoo Monkeypet chat bot’s 2016 US Presidential election day coverage was designed.</h4><p>While most people followed the 2016 US Presidential election day on their favorite news websites, search engines, TV &amp; radio channels, we at Yahoo experimented with covering that day as a conversation with a chat bot. This is the story of our experiment.</p><h3><strong>Why a chat bot?</strong></h3><p>I helped design the 2016 US Elections coverage for Yahoo Search.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2Fip5aO3MPJTIBVgDppRLg.png" /><figcaption>Designs for US Elections coverage on Yahoo Search</figcaption></figure><p>While researching for this project, I realized that a typical US Presidential Election Day has all the ingredients to make a thrilling story. In a matter of 9 to 12 hours, a country casts its votes, starts counting them, does a lot of math and analysis of possible results, sees the results swing one way to another and finally gets a winner.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/603/1*Z1tdBuyFKf73dJXLdVYtZg@2x.png" /><figcaption><em>Various states of an election day</em></figcaption></figure><p>A chat bot with its messages programmed to reflect the current state combined with live results coming from a dependable API would make it well equipped to live commentate such a day.</p><h3>My Role</h3><p>I designed an election results bot. Internally marketed it and then teamed up with the Yahoo Monkeypet team to integrate it in their bot.</p><h3>Use cases</h3><p>The project started by creating a list of questions an election results bot user may have and then categorizing them based on their underlying intent.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*I50Otlfp3ckegfsq6JA27Q.png" /><figcaption>What could election bot users want to know?</figcaption></figure><p>As the design evolved, it made sense to limit the bot’s focus to ‘Potentially most relevant to most people’ use cases like</p><blockquote>‘What is the current score?’<br>‘What does each candidate need to win from here?’<br>‘Who is winning which state, which states are still counting and which still voting?’</blockquote><h3><em>Navigation</em></h3><p>My first attempt to define navigation by choosing between a horizontal or a vertical flow failed as they could not capture the repetitive nature of a user’s engagement with such a bot.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NwxUxFOxIVcksqDf-Pa7vA@2x.png" /><figcaption><em>Failed navigation designs</em></figcaption></figure><p>I re-started by defining the conversational elements of each interaction between the person and the bot as …</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/180/1*vimp8HN6QDRDRsR2EST37A@2x.png" /><figcaption>Conversation structure of each user-bot interaction</figcaption></figure><p>… and then imagined what each of these elements could contain at the different stages of the election day. This exercise produced the first cut of what the bot could look like.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jKlPUhK1ISkvm6Izu-VQNw@2x.png" /><figcaption>First cut of bot design</figcaption></figure><h3>Refinement</h3><p>Multiple iterations were done on the first cut to increase information density of cards, demarcate what should be presented as cards and what as text and clarify the information hierarchy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/556/1*lxbFoZl5e_NRmj37MsPuIw@2x.png" /><figcaption>Text messages limited to showing status, summary and suggestions</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/546/1*zKpoD6J4Jysnnk_9U6mYRA@2x.png" /><figcaption>Final set of cards representing 3 levels of hierarchy</figcaption></figure><h3>Integration with Monkeypet</h3><p>We decided to integrate election results as a new skill of our already popular Monkeypet bot rather than to market and release a new bot just for a day.</p><p>Existing users were notified of this new skill and all bot dialogs were converted to Monkeypet speak <em>(by Monkeypet’s resident editor)</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/264/1*PphPLw698I_8a9slzPAP8w@2x.png" /><figcaption>Some of the new bot dialogs</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wRWRFwpT-WJiX6Lu9KQ7dw.png" /><figcaption>Final Flow</figcaption></figure><h3>How we performed</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/534/1*hknQZtH8I17iCXOVp4JbNg@2x.png" /><figcaption>Bot performance for election day vs an average day</figcaption></figure><p>We found people engaged with Monkeypet bot on election day many times more than they did on an average day. They responded to our election update notifications right till the winner was declared. We feel this to be a validation of our hypothesis that chat bots when designed right are a relevant medium for live coverage of significant events.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b6591b11128c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Rio Olympics on Yahoo search]]></title>
            <link>https://medium.com/@simarjeetsingh/rio-olympics-on-yahoo-search-ec4e77b20ca?source=rss-192e291131d9------2</link>
            <guid isPermaLink="false">https://medium.com/p/ec4e77b20ca</guid>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[user-experience-design]]></category>
            <category><![CDATA[olympics]]></category>
            <category><![CDATA[yahoo-search-design]]></category>
            <category><![CDATA[systems-thinking]]></category>
            <dc:creator><![CDATA[Simarjeet Singh]]></dc:creator>
            <pubDate>Fri, 30 Dec 2016 21:59:24 GMT</pubDate>
            <atom:updated>2017-09-13T22:48:29.843Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vZZmj2JifXpKi7hxN58MSQ@2x.png" /></figure><h4>How Yahoo search’s Olympic Schedule, Live Match, Medals and Results modules were designed.</h4><p>During the 2016 Olympic Games, Yahoo Search displayed information modules for olympics related searches like ‘Olympics’, ‘Team USA’ and ‘Michael Phelps’. These modules presented dynamic content like medal count, live events, upcoming schedule, recent results, latest pictures, videos and news. This case study is about my role in creating this experience.</p><h3><strong>My Role</strong></h3><p>We had 10 weeks to design, develop and launch our olympics experience. We divided our design effort into 8 main tracks. I was responsible for medals, results, live events &amp; schedules. The other design tracks were image &amp; video, icons, k<a href="https://en.wikipedia.org/wiki/Knowledge_Graph">nowledge graph</a> and news.</p><h3><strong>The Problem</strong></h3><p>We inherited a list of 40+ olympics use cases from the teams that worked on previous editions. Here are 2 such use cases:</p><blockquote>‘As a casual user, I want to search for a certain athlete that I don’t know of.’<br>‘As an Olympics fan, I want to search for Olympic results’</blockquote><p>I needed to define an approach to design for these use-cases. Should I tackle each use case separately? How would I map different modules to a use case? How would the modules change before, during and after the games or before, during and after an event?</p><h3><strong>My Approach</strong></h3><p>I created a matrix to capture how user needs would vary for olympic entities over time. This system clarified the number of flavors needed for each module and the inter-relationships between different modules.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/293/1*mkzbpe33VKYfTuqB5bSueg@2x.png" /><figcaption>Relevant modules for each entity-time combination</figcaption></figure><p>To design each module, I used an iterative process and evaluated numerous hypothesis and solutions.</p><h3><strong>Schedules</strong></h3><p>Multiple design explorations were done to define the optimum navigation, relevant information &amp; appropriate user actions.</p><p><strong>Navigation Explorations</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*ln6hvL1mxLA32ZS89FaUhQ@2x.png" /></figure><p><strong>Explorations to define relevant columns &amp; user actions</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*2BSQeDm3xW1JIooinB51qw@2x.png" /></figure><p><strong>Final Design</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/566/1*u2NJHDp9qLNcEQ-X4mUHqw@2x.png" /><figcaption>Final Template</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/958/1*LRc0NaHL_ICi62h4mOsnKg@2x.png" /><figcaption>A few schedule modules on desktop and mobile</figcaption></figure><h3>Live Events</h3><p>The main challenge in designing the live match modules was that while all sports are unique, we wanted a single consistent system that would work for them all. I defined my goals as:</p><blockquote>1. Design should be consistent across sports.<br>2.Design should give full snapshot of event.<br>3. Designed modules should work in tandem with schedule modules.</blockquote><p><strong>Initial Explorations</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*XVr92i3WCxKD7NPeQVDiew@2x.png" /></figure><p><strong>Final Design</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*OuzEw8e_ez3YjZOQss6V2A@2x.png" /><figcaption>Final Template</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/570/1*JhtLrQakTXEG10Jp9INwMw@2x.png" /><figcaption>The design works for multiple sports without compromising depth of information.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/954/1*Po74o1I7RJAAnA6o3F1PJA@2x.png" /><figcaption>The design works in tandem with the schedule modules on desktop and mobile.</figcaption></figure><h3>Results &amp; Medals</h3><p>Results &amp; Medals are very closely related modules. They had to be designed together. I had to carefully map when I needed both Medal &amp; Result module to fully answer a search query, when only Medals would be enough and when only Results would be enough. Here are a few examples:</p><p><strong>Both Results &amp; Medals</strong><em><br></em>Consider search term ‘Olympics’. Medals table gives a cumulative total of medals won &amp; country wise standings. It cannot highlight recent medals that just got added to this total. An additional ‘Latest results’ module gives that information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*KhsS4UH1Kj3dUCDvWDW7rg@2x.png" /></figure><p><strong>Only Medals</strong><em><br></em>Consider search term ‘Kayak’. The Medals table shows events that are over. Schedule module shows when and at what stage the remaining events are. We don’t need a separate results module.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*YQV-zWI1a8McYzX-xmIbzQ@2x.png" /></figure><p><strong>Only Results<br></strong>Consider search term ‘Michael Phelps’.<strong> </strong>During his events, the Results Module shows Michael Phelps’ progress through his events. As his events start getting over, the module starts showing his final position in those events.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*qf321X7gU3wasYqOzNW8bg@2x.png" /><figcaption>Results before any of the events is over</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*geuisQCGcU3XkvpMrqzLBQ@2x.png" /><figcaption>Results after some of the events are over</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*SKXXE2mOD2S89gMlNnH_zQ@2x.png" /><figcaption>Results after all events are over</figcaption></figure><h3>What people saw</h3><p>Medal, Results, Schedule and Live Modules came together with other content to answer people’s queries.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pknJbmgsgh-_Dl5UQvvThw@2x.png" /><figcaption>Search experiences for some popular queries during the olympics</figcaption></figure><h3><strong>How we performed</strong></h3><p>Our goal was to increase user engagement through better design. We measured how users interacted with our olympics modules in comparison to other current search user experiences as well as previous olympics experiences. Our performance is encapsulated below <em>(numbers and sensitive information withheld).</em></p><blockquote>1. Increase in time spent, clicks and re-queries.<br>2. Increase in subsequent engagement with Yahoo search.<br>3. Increase in volume of traffic sent to Yahoo Sports.<br>4. Decrease in query reformulations due to poor search results.<br>5. Olympics modules shown for more query types than ever before.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ec4e77b20ca" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>