<?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 Fei Fan | UX &amp; Product Design on Medium]]></title>
        <description><![CDATA[Stories by Fei Fan | UX &amp; Product Design on Medium]]></description>
        <link>https://medium.com/@iancheng0323?source=rss-90b397d3ad00------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*gF2vpPDG3mNW9oCyaS2MeA.jpeg</url>
            <title>Stories by Fei Fan | UX &amp;amp; Product Design on Medium</title>
            <link>https://medium.com/@iancheng0323?source=rss-90b397d3ad00------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 18:03:51 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@iancheng0323/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[如何把ChatGPT變成客製化的設計夥伴？]]></title>
            <link>https://iancheng0323.medium.com/%E5%A6%82%E4%BD%95%E6%8A%8Achatgpt%E8%AE%8A%E6%88%90%E5%AE%A2%E8%A3%BD%E5%8C%96%E7%9A%84%E8%A8%AD%E8%A8%88%E5%A4%A5%E4%BC%B4-28cfb94f7989?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/28cfb94f7989</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[ux-design]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Wed, 10 Jun 2026 17:28:30 GMT</pubDate>
            <atom:updated>2026-06-10T17:28:30.616Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ujmgoCRtFG-PAJaG" /><figcaption>Photo by <a href="https://unsplash.com/@thisisengineering?utm_source=medium&amp;utm_medium=referral">ThisisEngineering</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>你有和AI討論過你的設計嗎？體驗如何呢？</p><p>大多數時候，如果你給AI看你的設計，他都只能給出泛化且「AI感」十足的建議，對於實際的工作並沒有真實的幫助。</p><p>但我發現其實我們可以透過組合不同的功能，大幅提升與AI討論設計的品質，甚至可以到像是設計夥伴的程度。我說的不只是給出一些方向性的建議或是a11y審查那麼簡單，我是說讓AI可以產出專業且針對性你的產品的建議，就像一個資深設計師坐在你旁邊一樣。</p><h3>為什麼使用AI來Audit設計?</h3><p>我始終覺得廣納觀點可以改善設計品質，但這有時候並不容易。在小團隊中，你可能缺少能資深設計師來給建議。而在大公司裡，其他設計師或是設計主管也有自己的職責，不可能參與設計的各個階段，所以讓AI來審閱設計就變成最務實的做法。</p><p>雖然講起來很合理，但初期的AI理解UI的能力並不好，給出的設計建議品質也低，因此我們的團隊一開始也只利用AI來處理文件。</p><p>好在過去的幾個月裡，AI辨識UI意圖的能力大有進步 — 除了可以很快辨識出頁面上所有的文字與結構，也能針對不同設計階段產生針對的建議。例如丟一張Wireframe 給他，他會知道這不是黑白的設計；給他一張介面的兩個版本，他也能知道我需要他比較兩版設計。</p><h3>第一階段： Cutome Prompt 提升效率</h3><p>一開始，我們只是使用了Prompt去包裹設計並且交代Context，像是</p><pre>「幫我檢視這個設計介面，並且給我建議」</pre><p>但每次都要打出一樣的指令，這種小小的摩擦最後就會造成設計上的日常工作上的棄用，尤其在天天都用的工具。</p><p>我初期的解決方法是用Custome Instruction的方式讓AI知道，如果我貼了一張設計截圖進來，那麼他需要直接開始幫我們分析介面。</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/980/1*W4xMlHB9qwwi5tWfDVQfRw.png" /></figure><p>這個做法不只讓我節省了時間，也可以給他更多Context，例如他的回應風格、分析重點等等，讓建議與想法更有針對性。</p><h3>第二階段：Skills 提升品質</h3><p>但我很快發現Custom Instruction並不夠用，主要是因為AI的回答品質還是不夠高。</p><p>每個設計專案都有自己的設計目標，每個頁面也是，產品中也有許多功能與流程細節。為了提高回覆品質，我發現自己仍然需要讓AI知道這些。例如說我們是矯正眼鏡電商，因此會有選擇鏡片這一特別的流程。少了這些知識，AI就不容易產出高品質的回答。</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oT9n4duSLii4mJyHVK4oWA.png" /><figcaption>Lens Selection Process</figcaption></figure><p>但問題是 Custom Instruction 的空間有限，<a href="https://help.openai.com/en/articles/8096356-chatgpt-custom-instructions">而且也不適合用來儲存這些資訊</a>。（Custom Instruction 是用來讓AI了解與用戶應對的原則，不應該關於特定任務且有1500 字的長度上限）</p><p>因此，我們把這個這個指令包裝成一個 Skill。</p><p><a href="https://openai.com/academy/skills/">Using skills</a></p><p>Skill 的優勢是擁有自己的檔案系統，可以讓我加入多個參照檔案，這代表我不需要仰賴Custom Instruction，而且可以載入大量的資料。我用這個結構來告訴AI我們公司的基本資料、頁面定義與目標，以及我們在設計時最在意什麼。例如：</p><pre>PDP (product detail)<br>  Purpose: help an undecided buyer build confidence in one model.<br>  Primary goal: add-to-config rate. Secondary: spec comprehension.<br><br>Cart / Summary<br>  Purpose: confirm the build and surface trustworthy upsells.<br>  Primary goal: checkout start. Avoid: last-minute price shock.</pre><p>使用Skill的另外一個好處是它可以標準化回應時機、方式以及格式。例如我在Skill中明確地告訴 ChatGPT 需要先寫出正在Audit的頁面，並列出設計中的好與壞。</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*B4vOjDuJ-M6PvYrobBUvCg.png" /></figure><p>有了這樣的配置後，我可以很明顯地感覺到回答的品質有所提升。回覆格式統一之後，我們團隊也更能夠記錄與本此分享這些習得的資訊。</p><h3>第三階段: 利用Audit Context增加透明度</h3><p>這時候，我想我們已經有一個很很實用的Design Critique Buddy了，但還是有一個問題，那就是這個系統並不透明。</p><p>雖然AI本身就是一個大黑盒子，但要用在工作上，我們希望他越透明越穩定越好。為了達到這個目標，我們又在Skill的基礎上規定AI需要先產生一份Audit Context才能開始審閱與討論。</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_klKmI05on0YNtPncV4Fcw.png" /></figure><p>這樣一來，我們除了可以自己檢視AI對專案的理解，也可以保證AI在同一次的Audit中都會使用一樣的標準。</p><h3>結尾</h3><p>希望這個工具對你有用，如果有任何問題歡迎讓我知道。</p><p>謝謝你閱讀到這裡，如果覺得內容有用的話請幫我：</p><p>👏 拍手：1–50下都可以，根據你的喜好程度。<br>🔔 訂閱：Medium, <a href="https://www.youtube.com/channel/UCHoYQ1d2OANaoHU-1HQ9iIQ">Youtube</a> 還有 <a href="https://www.instagram.com/feifan.design/">Instagram</a><br>🕊️ 分享：給可能感興趣的人</p><p>我是 Fei Fan，在上海工作的 UX 設計師，我會定期以簡單的方式分享 UX 設計與產品管理的相關工具與技巧。</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=28cfb94f7989" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Think Like a CRO Specialist (Even If You’re Just a Designer)]]></title>
            <link>https://iancheng0323.medium.com/how-to-think-like-a-cro-specialist-even-if-youre-just-a-designer-af2f079b3e99?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/af2f079b3e99</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[cro]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[conversion-rate]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Thu, 23 Apr 2026 18:00:48 GMT</pubDate>
            <atom:updated>2026-04-23T18:00:48.642Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ucCy_P0EWrsuaWJx" /><figcaption>Photo by <a href="https://unsplash.com/@lukechesser?utm_source=medium&amp;utm_medium=referral">Luke Chesser</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Most CRO work I’ve seen is expensive guessing with a spreadsheet attached — and I say that having worked in this space for a long time, not as an outside criticism.</p><p>The tools are fine. The people are usually smart.</p><p>What tends to go wrong is the question at the start of the process — or the fact that there isn’t one.</p><p>I recognized this early in my career, before I even knew the CRO job title existed.</p><p>I’d always designed with conversion in mind, asking what was stopping users from moving forward and treating every design decision as something I’d need to validate against real behavior.</p><p>When I eventually came across the term <em>“CRO specialist,”</em> my reaction was less curiosity and more recognition — I’d been thinking this way for years without knowing it had a name.</p><h3>Why most CRO tests don’t teach you anything</h3><p>There’s a pattern I’ve noticed when CRO teams run out of strong ideas.</p><p>They start cycling through every variable they can reach — testing button placement against urgency copy against social proof positioning — running combinations until something moves.</p><p>And something usually <em>does</em> move, at least temporarily.</p><p>But because nobody stopped to think carefully about <strong>why that change would matter to a specific person at a specific moment</strong>, nobody can explain the result.</p><p>Which means nothing from that test actually informs the next one.</p><p>After a while, you end up with a long history of results that don’t add up to a coherent understanding of your users.</p><p>What I’ve found is that a real hypothesis needs a <strong>mechanism behind it</strong> — some actual understanding of what the user is experiencing that makes you believe a particular change will matter to them.</p><p>Without that, even when a test produces a clear result, you don’t really know what changed for the user.</p><p>And the next hypothesis ends up being just as arbitrary as the last.</p><h3>How I build one before I design anything</h3><p>The question I come back to before any design decision is:</p><p><strong>At this exact screen, for this exact user, given everything that brought them here — what is stopping them from taking the next step right now?</strong></p><p>Working through that question forces you to think about:</p><ul><li>what the user has already seen</li><li>what they’re worried about</li><li>how much patience they have left</li><li>what would need to be true for them to move forward</li></ul><p>Before I start designing, I want to be able to:</p><ul><li>describe the friction causing hesitation</li><li>say what I expect to change if I address it</li></ul><p>If I can’t do that clearly yet, it’s usually a sign: <strong>stay in the thinking phase.</strong></p><p>Don’t move into design and hope clarity shows up later.</p><h3>What this means if you’re a designer</h3><p>Designers are often better positioned for this kind of thinking than they realize. The ability to reason carefully about what a specific user is experiencing at a specific moment is genuinely hard to develop.</p><p>Most designers have spent years building exactly that. A lot of CRO practitioners haven’t.</p><p>If you can bring that foundation into how you approach conversion problems — and pair it with the discipline to really understand the user moment before you design — you’re already working the way the best people in this space work.</p><p>The shift that matters isn’t going from designer to CRO specialist. It’s going from:</p><p><strong>“what should this look like?”</strong></p><p>to:</p><p><strong>“what is stopping this person, right now, from taking the next step?”</strong></p><p>Hope this article helps, and happy designing!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=af2f079b3e99" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Generate Quality A/B Test ideas in 10 Minutes Using AI]]></title>
            <link>https://iancheng0323.medium.com/how-i-generate-quality-a-b-test-ideas-in-10-minutes-using-ai-eae52588bd32?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/eae52588bd32</guid>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[conversion-optimization]]></category>
            <category><![CDATA[ab-testing]]></category>
            <category><![CDATA[cro]]></category>
            <category><![CDATA[ideation]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Wed, 22 Apr 2026 17:47:47 GMT</pubDate>
            <atom:updated>2026-04-22T17:47:47.755Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jykicccnXkj6OZPi" /><figcaption>Photo by <a href="https://unsplash.com/@kimsuzi08?utm_source=medium&amp;utm_medium=referral">Suzi Kim</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Most articles about A/B testing spend a lot of time on statistical significance, sample sizes, and how to call a winner. I want to talk about the part that comes before all of that — the moment when you are staring at a blank page trying to decide what to test in the first place.</p><p>In my experience, that blank page is where most teams actually lose time, not in the running of the test itself.</p><p>I have been running A/B tests for a more than 10 years now, and the question I get asked most is not <em>“how do you analyze results”</em> but <em>“how do you come up with good ideas.”</em> I used to answer this with something vague about looking at analytics and heatmaps.</p><p>But over the last 2 year, I started working with AI and.developed a structured process to speed up the whole process — without losing quality.</p><h3>Step 1: Know Which Mode You Are In</h3><p>Before I open any AI tool, I ask myself one question:</p><p><strong>Am I trying to find the local optimum, or am I trying to switch to a different growth path entirely?</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*q5OL1UF9TU2Kjnw2" /><figcaption>Photo by <a href="https://unsplash.com/@sammgrdichian?utm_source=medium&amp;utm_medium=referral">Sam Mgrdichian</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>These are two different modes, and they require different kinds of hypotheses.</p><p><strong>Optimizing</strong> means working within what already exists — changing copy, adjusting colors, moving a CTA button. This kind of testing has a ceiling. You can push conversion up incrementally, but at some point you are squeezing the last few percentage points out of a design that has already been refined.</p><p><strong>Exploring</strong> means testing whether a fundamentally different approach — a new layout, a different flow, a new value proposition — opens a higher ceiling altogether.</p><p>Both are valid, but mixing them up wastes test cycles.</p><p>If I am in optimization mode but generate exploration hypotheses (or vice versa), most of the ideas will not actually fit the situation I am trying to solve.</p><p>Answering this question honestly before I start shapes:</p><ul><li>what I ask AI to do</li><li>what page context I provide</li><li>what kind of ideas I expect to come back with</li></ul><h3>Step 2: Prime the AI With Context</h3><p>Most people skip this step and then complain that AI gives generic output. The generic output is not the AI’s fault — it is the input. If you paste a URL and ask for <em>“ideas to improve conversion,”</em> you will get a list of things that could apply to any website on the internet.</p><p>What I do instead is keep the key infos like company background, customer profile, purchase flow, department objectives, key competitor context (basically the information you train your newby designer) stored in a personalization skill.</p><p><a href="https://help.openai.com/en/articles/20001066-skills-in-chatgpt">Skills in ChatGPT | OpenAI Help Center</a></p><p>Before I generate hypotheses for a specific page, the AI already knows:</p><ul><li>how our business operate</li><li>who our customers are</li><li>who we are up against</li><li>our current conversion priorities are</li></ul><p>That background changes the quality of the output significantly.</p><p>When I also add context about the specific user type I am designing for, the ideas get even more targeted.</p><p>This setup is a <strong>one-time investment</strong>. Once the personalization is in place, starting a new hypothesis session takes almost no extra time — the AI already has the foundation, and I just need to tell it the page and the goal.</p><h3>Step 3: Generate Iteratively, Filter by Logic</h3><p>The generation itself works best as a <strong>conversation</strong>, not a single prompt.</p><p>I give the AI:</p><ul><li>the target page</li><li>the conversion goal</li><li>my mode from Step 1</li></ul><p>Then I:</p><ol><li>Read the first batch of ideas</li><li>Pick the ones that feel promising</li><li>Tell the AI why I like them</li></ol><p>After two or three rounds of that feedback loop, the suggestions get noticeably more specific and more grounded. The AI starts to understand what kind of thinking I am doing.</p><p>Another approach I use is feeding <strong>screenshots of the design</strong> directly into the AI and asking it to audit the page before generating hypotheses. This forces the AI to engage with the actual visual details rather than making assumptions from a description alone.</p><p>The hypotheses that come out of this tend to have clearer logic behind them — especially if I also specify the user type. That step alone makes a noticeable difference.</p><h3>A Critical Warning About AI</h3><p>With this workflow, you will be able to produce handful of ideas in a blink. But you still need to be cautious, AI can still make mistakes after all the improvements. And worst of all, AI is <strong>very persuasive even when the ideas are bad</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*7W8IxxMT0NoMTSOL" /><figcaption>Photo by <a href="https://unsplash.com/@bmpanaguiton?utm_source=medium&amp;utm_medium=referral">Breana Panaguiton</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>It explains its reasoning confidently, and it is easy to approve something that sounds compelling in the abstract but falls apart the moment you see it designed.</p><p>I now always make a quick mockup of any hypothesis I am seriously considering, or.get a second opinion before committing to a test cycle.</p><p>These extra step has saved me from running tests that would have been fine in theory and confusing in practice.</p><h3>Putting It Together</h3><p>Knowing your mode, priming the AI, and running two or three rounds of iteration moves fast once the system is in place.</p><p>The validation step — mockups, second opinions, prioritization — takes longer, and it should.</p><p>That part is still a human judgment call, and it is where the real quality control happens.</p><p>What AI has done for me is remove the blank page entirely.</p><p>I go into every hypothesis session with:</p><ol><li>a clear mode</li><li>a well-primed context</li><li>a simple filter for evaluating ideas</li></ol><p>The output is not always brilliant, but it is always logical — and it always gives me something real to work with.</p><p><strong>Hope this article helps, and happy designing!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=eae52588bd32" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Most Designer Look at Competitors Analysis Wrong.]]></title>
            <link>https://iancheng0323.medium.com/most-designer-look-at-competitors-analysis-wrong-ed6d1ca1c2de?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/ed6d1ca1c2de</guid>
            <category><![CDATA[ux-research]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[competitor-analysis]]></category>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[design]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Tue, 21 Apr 2026 17:53:48 GMT</pubDate>
            <atom:updated>2026-04-21T17:53:48.364Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*b2s02sy2ne34izeV" /><figcaption>Photo by <a href="https://unsplash.com/@rocinante_11?utm_source=medium&amp;utm_medium=referral">Mick Haupt</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>Everyone Says ‘Just Look at Competitors.’ Most People Look at the Wrong Things.</h3><p>Whenever I talk to designers about competitive analysis, the process usually goes something like this: open a few competitors, map their features, screenshot their layouts, see what patterns emerge. I’ve done it on almost every project I’ve worked on. Most briefs include it. Most teams treat it as a standard first step.</p><p>And for a long time, I thought I was doing it well. I had comparison tables, organized screenshots, layout patterns neatly catalogued by category. It looked thorough. But looking back, most of what I was producing was just inventory — not analysis. I was noting what competitors had, not asking why they’d built any of it. One project made me realize how much that distinction matters.</p><h3>My experience — when I was just taking inventory</h3><p>I was working on a redesign of our coupon experience, and I started the way I always did: open five competitors and map what they’re doing.</p><p>One of them had moved the coupon input field from the cart page to checkout. My first reaction was to note it and move on. Seems like a quirk. Probably not relevant. I almost filed it under “interesting” and forgot about it — which, I’d come to realize, was exactly the kind of competitive analysis I’d been doing all along.</p><p>But I kept coming back to it. Why would a well-resourced team make this specific move? Moving the coupon field further down the funnel seemed like it would frustrate users who expect to find discounts before committing to checkout. There had to be a reason — I just hadn’t looked for one.</p><p>When I finally sat with it, the logic became clear. Moving the coupon to checkout mirrors how it actually works offline — you hand over your coupon at the register, not while you’re still browsing the store. It also simplified the cart to a single job: confirm what’s in your basket, maybe add something. And because this competitor had a loyalty points system that activated at checkout, the coupon integrated naturally into a unified payment step. It wasn’t a quirky UI decision. It was a coherent design philosophy, built for a specific kind of shopper who expects to manage all their discounts and rewards in one place.</p><p>That feature placement was a window into their entire positioning strategy. I’d almost walked straight past it because I was looking at what was there, not asking why it was there. That’s the wrong thing to be looking at — and I was doing it constantly.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*3M5uMBUIpjABWUg3" /><figcaption>Photo by <a href="https://unsplash.com/@covasoftware?utm_source=medium&amp;utm_medium=referral">Cova Software</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>Why most competitive research stays at the surface</h3><p>The problem, in my experience, isn’t that designers skip competitor research. It’s that most of us go in without a specific question.</p><p>When I don’t have a clear question, I default to documentation: catalogue everything, compare every pattern, note every layout choice. It feels thorough. But what I end up with is a lot of observations and very few insights. I know what exists in the market. I have no real idea what any of it means.</p><p>Every design decision a competitor makes — what to include, what to remove, what to combine, what to subtly de-emphasize — reflects a bet. A bet on what a specific user needs to accomplish a specific job.</p><p>When I started thinking about competitive design this way, the whole exercise changed for me. The question stopped being “Do they have this feature?” but became: “What does this choice tell me about who they’re designing for, and why did they decide it this way?”</p><h3>How I read competitor design now</h3><p>Here’s what my process looks like now, whether I’m doing broad research across a whole product or going deep on a specific feature.</p><p>Three questions, applied consistently — whether I’m doing broad research across the whole product or deep-diving on a specific feature.</p><h4>1. I start by deciding what question I’m trying to answer</h4><p>Before I open any competitor’s site, I ask myself: what specific design decision am I trying to inform?</p><p>Not:</p><ul><li>“What are competitors doing with their coupon experience?”</li></ul><p>But:</p><ul><li>“Should our coupon field live in cart or at checkout, and what’s the user reason?”</li></ul><p>A clear question turns what I observe into evidence. Without one, I’m just collecting screenshots.</p><h4>2. For each thing I observe, I ask what job it’s doing and for whom</h4><p>I’ve started training myself not to just note what’s there. When I see a design decision — especially an unusual one — I ask:</p><ul><li>What problem is this solving?</li><li>What kind of user would run into that problem?</li><li>Does that user sound like mine?</li></ul><p>I won’t always know for certain. I’m reverse-engineering their positioning from visible signals, not reading their internal product spec. But even an approximate answer is far more useful than a screenshot.</p><p>“This looks designed for users who are price-sensitive and actively comparison-shopping” gives me something to work with. A screenshot of their cart page, by itself, gives me almost nothing.</p><h4>3. I map what I’ve learned back to my own users before deciding anything</h4><p>Once I have a read on who a competitor seems to be designing for, I ask myself:</p><ul><li>Does the same job apply to my users?</li><li>If yes — does the same solution work for them?</li><li>Or do they need a different path to the same outcome?</li></ul><p>This is the step where research actually connects to real decisions.</p><p>Not just:</p><ul><li>“What did competitors do?”</li></ul><p>But:</p><ul><li>“Of everything I observed, which of their bets are worth placing myself — and which ones reflect a positioning I’m not trying to replicate?”</li></ul><h3>What to do when you don’t have answers</h3><p>When I share this way of working, the most common reaction I get is: “I don’t really know my competitors’ target users. I’d just be guessing.”</p><p>My honest answer: yes, you’d be guessing — and I think you should do it anyway.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*4D7vZCZbNmUq5Bkr" /><figcaption>Photo by <a href="https://unsplash.com/@beccatapert?utm_source=medium&amp;utm_medium=referral">Becca Tapert</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>I’m never going to have certainty about a competitor’s internal research. But what I’ve found is that the goal isn’t certainty. The goal is to ask a better question than “do they have this feature?”</p><p>When I try to infer who something was designed for — even imperfectly, even when I get it wrong — I’m forced to think about design as a series of deliberate choices rather than a collection of features. That reasoning process ends up being more valuable than whatever conclusion I land on.</p><p>You may never have a complete picture of who your competitors are designing for. That’s fine. The habit of asking is what separates a feature list from a point of view.</p><h3>Final thought</h3><p>The next time I sit down to do competitive research, I try to start with a different question.</p><p>Not:</p><ul><li>What do they have?</li></ul><p>But:</p><ul><li>What did they decide — and who did they decide it for?</li></ul><p>Features can be reproduced without really understanding them. Decisions require you to understand the reasoning first.</p><p>When I started treating competitor design as a set of deliberate bets on users and positioning, my research stopped being a checklist and started being something I actually learned from.</p><p>Hope this article helps, and happy designing!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ed6d1ca1c2de" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Your First Year of Management Will Probably Feel Off]]></title>
            <link>https://iancheng0323.medium.com/why-your-first-year-of-management-will-probably-feel-off-9549c06984d0?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/9549c06984d0</guid>
            <category><![CDATA[design-management]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[management]]></category>
            <category><![CDATA[product-manager]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Mon, 20 Apr 2026 18:03:03 GMT</pubDate>
            <atom:updated>2026-04-20T18:03:03.135Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*SEKmZX_YAXzJB1xw" /><figcaption>Photo by <a href="https://unsplash.com/@matthewhenry?utm_source=medium&amp;utm_medium=referral">Matthew Henry</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>When designers think about moving into management, the conversation usually goes the same way — less time to design, more time in meetings. All of that is real, and worth preparing for — but it’s not what actually derails most people.</p><p>However, I came to realize what catches most new managers off guard could be a subtler, and most of them don’t see it coming until they’re already deep in it.</p><p>At least that’s what happened to me.</p><h3>My First Year as a Manager</h3><p>After my promotion, the calendar filled up fast, but that part felt manageable — more communication, more coordination, all things I had expected.</p><p>I loved it at first. I felt more productive, having more impact, and the taste of power turned out to be a real thing. But things turned sour quickly. The work started feeling less enjoyable, I stopped looking forward to it, and everything began to feel like drowning.</p><p>What I hadn’t expected was how much of my sense of self was tied to seeing my work ship and get used.</p><p>As a designer, the feedback loop is tight — you put something out, you watch it get used, a metric moves, and you feel it. That feeling of knowing something you built actually worked was a bigger part of why I loved the job than I had ever stopped to examine.</p><p>As a manager, that signal disappears almost entirely.</p><p>The conversations you’re shaping, the decisions you’re influencing, the team dynamics you’re trying to build — all of it lives somewhere in the future, and you can go weeks without anything concrete to point to.</p><p>It was a painful process, but the shift that finally worked for me was simple to describe and hard to actually do: <strong>I stopped tracking what I made, and started tracking what my team shipped.</strong></p><h3>What happened when you’re promoted</h3><p>You were in meetings. You reviewed work. You had conversations that felt important but left no artifact. None of that looks like “making” in the way design does — you can’t point to a screen and say “that’s mine.”</p><p>As a manager, asking <em>“what did I make today?”</em> almost always returns an empty answer. Over time, that emptiness accumulates.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*dbjBvJffAk1Hi-tg" /><figcaption>Photo by <a href="https://unsplash.com/@enginakyurt?utm_source=medium&amp;utm_medium=referral">engin akyurt</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>It’s not that you’re not working. It’s that the measuring stick you spent years building your professional identity around has stopped fitting the job.</p><h3>My team’s output is my design</h3><p>What eventually shifted things was a realization that <strong>my team’s output wasn’t separate from my design work. It was an extension of it.</strong></p><p>When I shaped how a designer framed a problem, when I pushed back on a brief before it reached the team, when I asked the right question in a review — my thinking was in that work. It just didn’t carry my name.</p><p>I had spent years building my craft, and a big part of my identity was tied up in being the person who made the thing.</p><p>Letting that go — genuinely letting it go, not just telling myself I should — turned out to be its own kind of work.</p><h3>The math that made it click</h3><p>What finally convinced me was the math. On my own, I could only ever work on one thing at a time.</p><p>My team could run experiments in parallel, explore multiple directions at once, and ship more in a month than I could have managed alone in a year. Every hour I spent trying to stay relevant as an individual contributor was an hour I wasn’t putting into that leverage.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*UQGH8yyaGTreeu_O" /><figcaption>Photo by <a href="https://unsplash.com/@nataliepedigo?utm_source=medium&amp;utm_medium=referral">Natalie Pedigo</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>I started tracking my team’s output the same way I used to track my own — what shipped, what moved, what got better. And for the first time, I had real answers at the end of each day.</p><p>The disorientation I felt in that first year wasn’t a sign that I was the wrong person for the role. It was a sign I was measuring myself with the wrong tool.</p><p>Once I switched tools, the job started making sense — and more importantly, it started feeling worth doing again.</p><p>Looking back, moving from designer to design manager was less a promotion than a change of medium.</p><p>My feedback loops got longer, my wins became less visible, and almost every tool I’d developed for measuring my own quality stopped working.</p><p>If you’re in that disorienting place right now, I don’t think it means you’re the wrong person for the role. It probably just means you’re still holding the old measuring stick. <strong>The job starts to make a different kind of sense once you put it down.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9549c06984d0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to tell you don’t have enough information in your UI?]]></title>
            <link>https://iancheng0323.medium.com/how-to-tell-you-dont-have-enough-information-in-your-ui-14002d1949b0?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/14002d1949b0</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[ui-design]]></category>
            <category><![CDATA[design-methodology]]></category>
            <category><![CDATA[ux-design]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Sun, 19 Apr 2026 17:10:55 GMT</pubDate>
            <atom:updated>2026-04-19T17:10:55.328Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bXkSg8lPn4ov5fXR" /><figcaption>Photo by <a href="https://unsplash.com/@theesplanadeshop?utm_source=medium&amp;utm_medium=referral">Andrew Dunstan</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>We know what happens when a page overwhelms users with too much information — they freeze, skim without reading, and leave. Most teams have already internalized the instinct to simplify and cut.</p><p>But there’s a quieter failure that’s just as damaging and much harder to spot: <strong>the page that doesn’t say enough</strong>.</p><h3>Why it’s hard to spot</h3><p>Too much information and not enough information produce the same surface-level result: <strong>users don’t convert</strong>. So when teams see drop-offs, the default reaction is to simplify further — remove content, shorten pages, reduce options. That works when the problem is overload, but when the issue is missing information, it makes things worse.</p><p>Users don’t announce it or show visible frustration — they just leave, which makes it easy to assume the UI is already clean and that something else is causing the drop-off.</p><h3>The change that looked right but wasn’t</h3><p>I saw this play out while working on a lens configuration flow.</p><p>The logic seemed solid: once users are deep in a step-by-step purchase configurator, they don’t need distractions like sitewide promotions. At that stage, they should be focused on making decisions, not evaluating deals.</p><p>We removed the promotion banner — and conversion dropped. What we missed was simple in hindsight: lenses make up a significant portion of the total order cost, and for many users, the promotion wasn’t a distraction — it was part of the decision. It answered the question they were already asking: ‘Is this worth it? Am I getting a good deal here?’</p><p>We thought we were reducing noise. What we were actually removing was information users needed to justify moving forward.</p><h3>How I approach it now</h3><p>Instead of thinking in terms of more or less, I approach this as a decision problem. Every screen exists to help users make one specific decision — and if you’re not clear on what that decision is, it’s almost impossible to know what information belongs there. The process I use breaks down into three steps:</p><ol><li><strong>Define the decision</strong></li><li><strong>Identify the blocking question</strong></li><li><strong>Map the minimum information needed</strong></li></ol><h4>1. Define the decision</h4><p>This is where most ambiguity actually starts — and it sounds more obvious than it is. Teams typically describe screens in terms of content categories or features: ‘this page explains our options,’ ‘this step shows product details.’ But users aren’t thinking about content. They’re trying to answer a specific question so they can move forward.</p><p>A more useful frame: <strong>what decision should the user be able to make after this screen?</strong></p><p>For a lens configuration step, that might be ‘choose the right lens package,’ ‘decide whether to continue to checkout,’ or ‘pick between two options.’ If you can’t describe the decision in one sentence, the screen will drift — over-explaining in some areas, under-explaining in others, with no clear anchor for what matters.</p><h4>2. Identify user types and blocking questions</h4><p>Once the decision is clear, the next step is understanding what might stop someone from making it. This is where different users start to matter — not in the abstract sense, but in what they specifically need in order to act.</p><p>A first-time user is still trying to understand: ‘What does this mean? Can I trust this?’ A user who’s already comparing options is asking: ‘Which one is right for me?’ A high-intent user who’s close to committing wants to know: ‘Is this worth it? What’s the total cost?’ Same screen — three different blocking questions.</p><p>Most designs go wrong by trying to serve all three equally, which leads to scattered information that doesn’t fully resolve any one concern.</p><p>The more useful goal is to<strong> identify who the primary user is</strong> at this specific step and what is the one question they most need answered.</p><p>Each user type has a different kind of block: uncertainty (they don’t understand), mismatch (they’re not sure it fits), or risk (they’re not sure it’s worth it). Address the dominant one, and you move users forward. Try to address all three at once, and you dilute everything.</p><h4>3. Map the minimum information needed</h4><p>Once you know the blocking question, the job becomes much simpler — and much stricter. Instead of asking ‘Is this information useful?’ you’re asking ‘Does this directly help answer the blocking question?’ For every element on the page:</p><ul><li>If it reduces the blocker → it earns its place</li><li>If it adds context but doesn’t help decide → it competes for attention</li></ul><p>UIs go wrong here all the time — not because they include bad information, but because they include correct, even helpful information at the wrong moment, in too much volume, or with the wrong framing. The goal isn’t completeness — it’s making sure the piece that unlocks the next step is clear, visible, and impossible to miss.</p><p>Most information problems aren’t really about quantity — they’re about screens built without a clear answer to what decision they need to enable. Get that right first, and what belongs and what doesn’t usually becomes obvious.</p><p>Hope this article helps, and happy designing!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=14002d1949b0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why business team doesn’t seem to care about UX?]]></title>
            <link>https://iancheng0323.medium.com/why-business-team-doesnt-seem-to-care-about-ux-3984b0c3987c?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/3984b0c3987c</guid>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[design-leadership]]></category>
            <category><![CDATA[ux-design]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Sat, 18 Apr 2026 15:05:05 GMT</pubDate>
            <atom:updated>2026-04-18T15:05:05.585Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*lXEHRYFKthT0zymf" /><figcaption>Photo by <a href="https://unsplash.com/@arnosenoner?utm_source=medium&amp;utm_medium=referral">Arno Senoner</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Have you ever felt ignored while giving design recommendations, despite being the only designer in the room? I sure have.</p><p>For a long time, I thought it was a power problem — the stakeholder had more authority, the organization didn’t prioritize design, or they just don’t care about UX. That framing made sense, but it kept me stuck because power dynamics aren’t something I could fix on my own.</p><h3>Why does this happen?</h3><p>Most stakeholders do care about user experience, at least in principle. The real issue is that <strong>“better UX” doesn’t connect to anything they’re measured on day-to-day.</strong></p><p>Everyone’s accountable to something — revenue, retention, cost, risk reduction. A recommendation that doesn’t mention any of those things is hard to act on, even if they agree with you.</p><p>Your PM is measured on conversion and feature adoption. Your engineering lead is optimizing for velocity and technical debt. Your executive sponsor cares about growth metrics and competitive positioning.</p><p>They’re not against good UX — <strong>they just need to see the chain of causality from user behavior to business outcome</strong>. Without that connection, your recommendation sits in a backlog next to dozens of other “good ideas” that never ship.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*c-Tjb8dM1xtn_ItR" /><figcaption>Photo by <a href="https://unsplash.com/@silverkblack?utm_source=medium&amp;utm_medium=referral">Vitaly Gariev</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>In most cases, you’re not dealing with someone who disagrees — you’re just speaking different languages. You’re framing the experience; they’re looking for the business case. Once that clicked for me, I started preparing recommendations differently.</p><h3>How I communicate UX now?</h3><p>I’ve found that structuring recommendations around three questions consistently gets better traction:</p><ol><li><strong>What user behavior is broken or missing?’</strong></li><li><strong>What does that behavior cost the business?</strong></li><li><strong>What’s the smallest change that would fix it?</strong></li></ol><p>These questions map directly to how stakeholders evaluate proposals: defining the problem, quantifying the impact, and assessing the solution.</p><h4><strong>1. What user behavior is broken or missing?</strong></h4><p>Specificity matters here. Instead of “the UX is unclear,” frame it as “users are dropping off at step <strong>X at a rate of Y%</strong>.” Concrete behavioral observations, <strong>especially with numbers</strong>, make the problem tangible and harder to deprioritize. Vague statements leave room for interpretation and delay.</p><h4><strong>2. What does that behavior cost the business?</strong></h4><p>Translate the user problem into metrics the organization already tracks — abandoned sessions, conversion loss or retention impact. Framing it as a <strong>conversion opportunity rather than a usability issue</strong> moved it from backlog to roadmap.</p><h4><strong>3. What’s the smallest change that would fix it?</strong></h4><p>Propose a specific, testable intervention rather than a comprehensive redesign. Smaller, well-defined asks reduce implementation risk and make it <strong>easier for stakeholders to commit</strong>. You’re asking for a decision on one change, not buy-in for a vision.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*OzMJLySTjHsALusU" /><figcaption>Photo by <a href="https://unsplash.com/@purzlbaum?utm_source=medium&amp;utm_medium=referral">Claudio Schwarz</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>When you speak their language, priorities shift</h3><p>Designers who consistently get recommendations approved aren’t necessarily producing superior work — they’re investing in translation. They connect observed user experience issues to the metrics and outcomes their organization already monitors.</p><p>This doesn’t require becoming a business analyst. It requires asking one additional question before presenting: <em>What does it cost if we don’t fix this?</em> That single shift typically reframes the entire conversation.</p><p>Hope this article helps, and happy designing!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3984b0c3987c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Avoid Analysis Paralysis in the Age of AI (As a Designer)]]></title>
            <link>https://iancheng0323.medium.com/how-i-avoid-analysis-paralysis-in-the-age-of-ai-as-a-designer-1b9bbda48740?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/1b9bbda48740</guid>
            <category><![CDATA[ai-design-tool]]></category>
            <category><![CDATA[design-process]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[productivity]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Fri, 17 Apr 2026 15:46:26 GMT</pubDate>
            <atom:updated>2026-04-17T15:46:26.799Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*tOO6ijnr5RGQ7QSC" /><figcaption>Photo by <a href="https://unsplash.com/@the_meaning_of_love?utm_source=medium&amp;utm_medium=referral">Aarón Blanco Tejedor</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Have your design process gone faster and easier with AI?</p><p>I have to admit I am on the side that did not really receive an efficiency boost.</p><p>It is not that AI did not help — it did. But it also made things harder to complete in every practical sense. The tools generate options and insights faster than teams can act on them. At the end of the day, not that much gets done.</p><p>How did this happen? Shouldn’t AI help us work faster with less effort?</p><h3>Why AI makes this worse, not better</h3><p>In design, as in any business context, uncertainty is uncomfortable. So we keep going and throw in more time and effort.</p><p>The twist with AI is that it removes the natural friction that used to slow you down in the past. When a competitive analysis that once took two days now takes ten minutes, there’s nothing stopping you from running three more. But beyond a threshold, more information stops helping and starts creating noise.</p><p>I’ve come to realize that recognizing when you’ve done enough — and having the discipline to stop — is one of the most underrated skills a designer can develop in the age of AI.</p><h3>How I decide when to stop?</h3><p>To prevent myself from falling into this trap, I now do the following:</p><h4>1. Decide what is good enough</h4><p>Every research type has a standard. In our team, for example, competitor analysis means looking at a minimum of 2 references; each design exploration usually needs to cover 3 variations. The bar sets both a minimum and an expectation — so that you neither under-do nor overdo.</p><p>Stop when you have passed the bar. Not when you feel certain — that moment rarely comes with design decisions. The standard you set in step one is your permission to stop analyzing and start deciding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*DV0X4dkh2bfpeHyN" /><figcaption>Photo by <a href="https://unsplash.com/@silverkblack?utm_source=medium&amp;utm_medium=referral">Vitaly Gariev</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h4>2. List down what you “can’t” know</h4><p>Sometimes the reason you cannot stop is that you are looking for an answer that research simply cannot give you.</p><p>There is only so much you can learn from references and competitive analysis. Some information can only come from testing — Does this layout feel right to the first time user? Will users notice this CTA?</p><p>Writing these questions down helps you acknoledge the limitation of research and also your blue print for learning.</p><h4>3. Time-frame your design process</h4><p>List down the milestones. Schedule checkpoints with your manager and stakeholders. Use these to establish discipline and push yourself forward.</p><p>Without a structured timeline, research has no natural end point.A checkpoint with your manager or a stakeholder can help you force a decision. When you know you have to present findings by Thursday, you stop generating on Wednesday.</p><p>Parkinson’s Law suggest that without a clear deadline, project takes forever to complete. This sounds dumb but is so true in the actual working space.</p><p><a href="https://asana.com/zh-tw/resources/parkinsons-law">Parkinson&#39;s Law: Why Work Expands &amp; How to Stop It Now [2025] * Asana</a></p><h3>Real obstacle: avoiding the decision</h3><p>Another real possibility is that you are using analysis to stall a decision. This is very common when information is cheap — gathering more feels productive, but the real move you are avoiding is committing to a direction. Being aware of that pattern is already half the battle.</p><p>Over-analyzing is tempting — and honestly, it makes sense. Design decisions get questioned constantly, and more data feels like protection.</p><p>With AI lowering the cost of analysis to nearly zero, that impulse is harder to resist than ever. If you have caught yourself running one more prompt at 11pm trying to feel ready, you are not doing something wrong. You are doing something very human.</p><h3>To wrap up</h3><p>Setting a bar, listing what you cannot know, and putting a timeline on your process are not constraints. They are what make a decision possible — and shipping is the only place where the real learning happens.</p><p>Hope this article helps, and happy designing!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1b9bbda48740" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Giving Users More Actually Makes Them Buy Less]]></title>
            <link>https://iancheng0323.medium.com/why-giving-users-more-actually-makes-them-buy-less-19f3cfcd0e68?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/19f3cfcd0e68</guid>
            <category><![CDATA[cro]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[decision-fatigue]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[analysis-paralysis]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Thu, 16 Apr 2026 06:59:26 GMT</pubDate>
            <atom:updated>2026-04-16T06:59:26.448Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*JURfXBPoeRHGHueP" /><figcaption>Photo by <a href="https://unsplash.com/@victoriano?utm_source=medium&amp;utm_medium=referral">Victoriano Izquierdo</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>The most common assumption in product design is that more information builds trust, and more trust drives conversion. If users aren’t buying, the thinking goes, <strong>they must need more proof</strong>.</p><p>In practice, that assumption breaks down past a certain point — <strong>more information on UI could actually negatively impact your conversion</strong>.</p><h3>My first hand experience</h3><p>A few years ago, we conducted an AB test adding more information to our product pages.</p><p>A version as control, while B version shows more specs, additional trust signals and the other product info.</p><p>The logic was that users weren’t converting because they needed more information and reassurance. However to our surprise, conversion went down significantly, with less conversion, less engagment and higher bounce.</p><h3>Why this happens?</h3><p>The brain doesn’t evaluate information neutrally when it’s under pressure to decide (like in your purchase flow). It reads <em>volume as complexity, and complexity as risk</em>. The safest response to perceived risk is to delay.</p><p>The jam study by Iyengar and Lepper showed this clearly: shoppers were 10x more likely to purchase when presented with 6 options rather than 24, with everything else held equal. The finding is usually framed as a lesson about choice, but the deeper point is about attention cost.</p><p>Every element a user processes raises the cognitive cost of deciding — and at some threshold, that cost exceeds the value of the purchase.</p><p><a href="https://cigdemgizemokkaoglu.substack.com/p/the-paradox-of-choice-jam-experiment">The Paradox of Choice: Jam Experiment and Its Application to Legal Services</a></p><h3>Different users, different fears</h3><p>Different users arrive at the same page carrying different blocking fears. Each person is at a different point in their decision journey — one is still deciding whether to trust you, another is deciding whether this is the right fit.</p><p>The implication is that the page’s job isn’t to present information — it’s to <strong>resolve the right fear for the right person at the right moment</strong>. That means organizing around visitor context, not content categories.</p><p>Take a software pricing page for example, A first-time visitor isn’t reading the feature matrix yet — they’re still deciding whether to trust the company enough to engage at all; An evaluator comparing tools already trusts the brand; they need the one detail that maps to their specific workflow.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dP9WHu4gfto3XFSm51BqCQ.png" /><figcaption>A Pricing Page</figcaption></figure><p>Same page. Same words. Everyone stuck for a completely different reason.</p><h3>How I decide what information to show?</h3><p>Here’s a diagnostic that I use that takes less than an hour:</p><ol><li>Identify the one decision you want users to make.</li><li>Name the single information most likely needed for that decision for your primary visitor.</li><li>Review every element on the page — does this address the blocking fear, or does it give users something else to process?</li></ol><p>Most of the time, the content isn’t wrong — it’s just in the wrong place, at the wrong volume, for the wrong visitor. And fixing these can drastically improve the conversion of the page.</p><p>If you are limited by time, pay extra attention to what sits above the fold. If the clearest path to conversion is buried below a comparison module, the page is working against you before users even scroll.</p><h3>Why this is hard to act on</h3><p>Cutting content is harder than adding it. Every element on a page has a champion, and removing it requires data, conviction, and usually a conversation nobody wants to have.</p><p>Instead of asking “Is this information useful?”, ask “Does this move users toward the decision, or does it just give them <em>more to think about</em>?”</p><p>Hope this article helps, and happy designing!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=19f3cfcd0e68" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Slowing Users Down Increases Conversions]]></title>
            <link>https://iancheng0323.medium.com/why-slowing-users-down-increases-conversions-cae9f9e50021?source=rss-90b397d3ad00------2</link>
            <guid isPermaLink="false">https://medium.com/p/cae9f9e50021</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[cro]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[conversion-rate]]></category>
            <dc:creator><![CDATA[Fei Fan | UX & Product Design]]></dc:creator>
            <pubDate>Wed, 15 Apr 2026 02:34:16 GMT</pubDate>
            <atom:updated>2026-04-15T02:36:50.237Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bnet8KkzQg5Smx6_" /><figcaption>Photo by <a href="https://unsplash.com/@anisagauri?utm_source=medium&amp;utm_medium=referral">Anisa Gauri</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>The instinct in UX work is to treat friction as something to minimize.</p><p>Extra steps add cognitive load, forms create drop-off, unnecessary clicks give users more opportunities to leave. This thinking is correct most of the time, and it’s one of the more reliable frameworks for improving conversion.</p><p>But not all friction belongs in that category. Some of it is doing something useful, and removing it without understanding its function can hurt more than it helps.</p><h3>My Experience</h3><p>Couple of years ago, we tested two versions of checkout flow in our company:</p><ul><li><strong>Version A:</strong> required users to choose between logging in or continuing as a guest before reaching checkout</li><li><strong>Version B:</strong> removed that step, taking users straight to checkout with a login option available but not required upfront.</li></ul><p>The hypothesis was that fewer steps (Version B) would lead to better conversion — one less decision, one less moment of hesitation before the purchase.</p><p>The results showed the opposite. Version B performed worse.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nP7-wjMI8RpeHpxF" /><figcaption>Photo by <a href="https://unsplash.com/@nathanareboucas?utm_source=medium&amp;utm_medium=referral">Nathana Rebouças</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>What we found was that users who would have logged in under Version A simply didn’t under Version B — they proceeded as guests, without logging in thus having lower intent to completed the purchase. Although more people entered the funnel, the funnel itself had become less effective.</p><p>That experiment changed how I think about friction. What looks like an obstacle is sometimes doing real work in the background — either filtering for higher-intent users or preparing users for what comes next.</p><p>So instead of asking “How do we remove friction?”, a better question is:<br>“What job is this friction doing?”</p><h3>Three situations where friction tends to improve outcomes</h3><h4>1. High-stakes, irreversible decisions</h4><p>When a user is about to make a choice that’s difficult to reverse, a moment of deliberate pause protects them from regret. Users who arrive at a commitment with more certainty tend to follow through more reliably, and they leave fewer complaints when they do. A confirmation step isn’t a UX problem — it’s a signal to the user that you want them to be sure.</p><p><strong>Example:</strong></p><ul><li>“Are you sure?” confirmations before deleting an account</li><li>Reviewing order detail before checkout</li></ul><h4>2. Filtering for intent</h4><p>Some friction points act as filters, selecting for users who are more invested in the outcome. Removing them often improves top-of-funnel volume while quietly degrading the quality of conversion downstream — which is exactly what happened in our checkout test.</p><p><strong>Example:</strong><br>Job applications that require a short written response or portfolio submission (instead of just a one-click “Apply”) consistently attract fewer applicants — but higher-quality ones.</p><h4>3. Building perceived value through effort</h4><p>Research found that people consistently value things more when they’ve invested effort in creating or configuring them — what they called the IKEA effect.</p><p><a href="https://medium.com/psychologs-magazine/the-ikea-effect-why-we-value-what-we-create-56ea8e63afdc">The IKEA Effect: Why We Value What We Create</a></p><p>Some experience like onboarding flows blocks the user from the product experience at first, but asking users to set goals or customize their experience tend to produce stronger early engagement because the user has already committed something.</p><p><strong>Example: </strong>Amazon still asks users to “clip” coupons instead of auto-applying them. That small action creates a sense of earning the deal, making the discount feel more valuable than if it were applied automatically.</p><p>Not every friction point belongs in the same category.</p><p>Some of it is creating unnecessary resistance that should be removed, some of it is doing something your product actually needs. The work is in telling the difference — and that usually starts with understanding what a step is for before deciding whether to cut it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cae9f9e50021" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>