<?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 Yang on Medium]]></title>
        <description><![CDATA[Stories by Fei Yang on Medium]]></description>
        <link>https://medium.com/@fyang5?source=rss-99f231f2cf4c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*yeagUH_lPA3x0pg6</url>
            <title>Stories by Fei Yang on Medium</title>
            <link>https://medium.com/@fyang5?source=rss-99f231f2cf4c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 19:23:08 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@fyang5/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[Before Misalignment Becomes Momentum]]></title>
            <link>https://medium.com/@fyang5/the-decision-layer-db8bd9912104?source=rss-99f231f2cf4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/db8bd9912104</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ux-design-leadership]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[design-strategy]]></category>
            <dc:creator><![CDATA[Fei Yang]]></dc:creator>
            <pubDate>Sat, 18 Apr 2026 15:57:50 GMT</pubDate>
            <atom:updated>2026-04-28T06:37:37.962Z</atom:updated>
            <content:encoded><![CDATA[<p>Design leadership in the age of AI-generated output</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aNSFDg0CnrGe_isNUfQ3QQ.png" /></figure><h3>I. The Great Unbundling of the Artifact</h3><p><a href="https://academics.design.ncsu.edu/student-publication/wp-content/uploads/2016/11/Simon_H_ScienceofArtificial.pdf">Herbert Simon</a> wrote that “everyone designs who devises courses of action aimed at changing existing situations into preferred ones.” That definition still matters because it keeps design honest: design is not simply about making things look better. It is about making situations work better for people.</p><p>AI has not changed that definition. It has changed what an artifact means inside that definition.</p><p>A designer can now produce, in an afternoon, what once took a team a week: interface variations, copy alternatives, prototype flows, research summaries, presentation structures, and system diagrams. The visible output of design has been separated from the time and effort it once required.</p><p>That is a significant shift. But if we follow the shift past the efficiency gain, the deeper implication is this:</p><p><strong>If machines can generate the “what,” then human designers must own the “why.”</strong></p><p>The artifact was never the most important thing designers produced. The most important thing was the judgment behind it: why this problem, why this framing, why this trade-off, why this direction and not the alternatives.</p><p>That judgment used to be partially hidden inside the artifact. A strong designer’s thinking could be inferred through the quality of the final work. But AI has made polished artifacts easier to generate. As a result, the logic behind the work has become more visible, and more important.</p><p>AI has not made design judgment less necessary. It has made weak judgment easier to scale.</p><p>A misframed problem that once produced one flawed prototype can now produce forty polished variations, all pointing in the wrong direction. The danger is not that AI produces bad-looking work. The danger is that AI can make weak logic look complete.</p><p>AI is a mirror. It reflects the quality of our thinking back at us quickly. If the framing is strong, AI helps us explore it faster. If the framing is thin, AI helps us produce thin work faster, with better typography.</p><p>That is why my focus has moved toward what I call the <strong>decision layer</strong>: the part of design work that defines what deserves to become an artifact in the first place.</p><p>For me, this perspective came through a two-year rebuilding period outside a conventional full-time design role. During that time, I was not only preparing to re-enter the field. I was studying how the field itself was changing. I ran a small online shop, returned to ArtCenter, studied EPD collaboration, researched companies like Stripe, and learned to work with AI not as a shortcut, but as a thinking partner that requires stronger human judgment, not weaker.</p><p>That period changed my understanding of design leadership. It is not simply about producing more. It is about helping teams decide better.</p><h3>II. When Artifacts Become Noise</h3><p>There is a failure mode that design teams rarely name: the proliferation of artifacts that creates the appearance of progress without the substance of alignment.</p><p>Figma files multiply. Decks accumulate. Workshop outputs are photographed and filed. Research documents sit in shared drives that no one returns to. The team has been productive in the narrow sense that objects have been produced. But the team has not necessarily been effective, because the objects do not share a decision logic that everyone understands.</p><p><a href="https://link.springer.com/article/10.1007/BF01405730">Rittel and Webber </a>described “wicked problems” as problems that resist definitive formulation, where solutions are not simply true or false, and where there is no clear stopping rule. Most meaningful product decisions have this quality. They do not have one correct answer. They require negotiation between users, business goals, technical constraints, risk, and time.</p><p>In that environment, more artifacts do not automatically create more clarity. They can create the opposite.</p><p>In one cross-functional product collaboration I worked on, the team had no shortage of material. We had prototype screens, workflow diagrams, feature ideas, research notes, and documentation of where communication had broken down. On the surface, the work looked productive. But the real issue was not whether we had enough artifacts. It was that we had not yet agreed on the exact decision moment the system was supposed to support.</p><p>Was the problem a communication gap? A handoff failure? A missing record of why a decision had been made? A lack of ownership over a recurring judgment call?</p><p>Each answer would lead to a different product direction. Until the team could name which answer it was operating from, the artifacts were premature. They were solutions in search of a confirmed problem.</p><p>This is what I mean by the artifact as liability. In an AI-augmented workflow, the liability compounds. If a team can generate twice the artifacts in the same amount of time, and those artifacts are not anchored to a shared decision logic, the noise scales with the speed.</p><p>Design leadership in this environment is not always about adding more. Sometimes it is about reducing the work to the decision that matters most.</p><p>Better collaboration is not more meetings. It is better decision architecture.</p><p>A useful decision architecture makes five things explicit:</p><ul><li><strong>What decision is actually being made</strong></li><li><strong>Who owns the decision</strong></li><li><strong>What trade-off the team is accepting</strong></li><li><strong>What uncertainty still remains</strong></li><li><strong>Where the reasoning will live so others can revisit it later</strong></li></ul><p>That last piece, organizational memory, is undervalued in many design cultures. A decision log is not documentation for its own sake. It is a record of how a team exercised judgment under uncertainty. When a direction is challenged later, the log answers not only what was decided, but why it was decided, what the team knew, and what it did not know at the time.</p><p>That is accountability infrastructure.</p><h3>III. Quality Is Operational</h3><p>Stripe is often discussed in design circles for its craft: the clarity of its documentation, the restraint of its interface, and the coherence of its product experience. What interests me most is not only the visible quality. It is the operating logic that seems to support that quality.</p><p>From the outside, Stripe is a useful public example of a company where craft appears connected to culture, standards, engineering partnership, and strong product judgment. Design leaders like <a href="https://www.lennysnewsletter.com/p/building-beautiful-products-with">Katie Dill</a> have spoken publicly about design quality, high-performing teams, and the relationship between craft and product-building. The lesson I take from Stripe is not that every company should copy Stripe’s interface. It is that quality is a system.</p><p>Quality does not happen only at the end of a process, during the polish pass before launch. It is built into how a team decides what to make, how it defines “done,” how it reviews work, how it handles the tension between speed and care, and how it protects the user experience from business decisions made without design judgment in the room.</p><p>In the AI era, if output becomes cheap, quality cannot depend on output review alone. It has to live upstream, in the operating culture that shapes what gets made and why.</p><p>Quality becomes operational, or it becomes accidental.</p><p>This principle applies at every scale. It matters in a company the size of Stripe, and it also matters in a one-person shop. The scale changes, but the underlying truth does not: trust depends on the operating choices around the product, not only the product itself.</p><h3>IV. Trust Is Designed Before It Is Measured</h3><p>Running a small Etsy shop taught me something about product experience that I have not been able to unlearn.</p><p>In a small business, trust is not theoretical. It is built or weakened through specific details: the accuracy of a product photo, the clarity of a shipping expectation, the tone of a message when something goes wrong, the care guide included in the package, and the response to a review.</p><p>A customer can receive exactly what they ordered and still walk away uncertain. Not because the product failed, but because the information around it failed to set the right expectation.</p><p>That taught me a precise thing about UX: <strong>the product is never only the object. It is the expectation architecture around the object.</strong></p><p>In software, the same principle appears at every scale. A user does not interact only with a screen. They interact with what the screen helps them understand, trust, and decide. Unclear status messages, hidden consequences, ambiguous warnings, and confirmations that confirm nothing are not minor UX issues. They are trust debt: small withdrawals from the user’s confidence that accumulate quietly until a moment of real consequence makes them visible.</p><p>The implication for design leadership is that trust cannot be treated only as a lagging indicator, measured after users complain, churn, or contact support. Trust is designed into the system from the beginning, in decisions about what information to surface, what consequences to make explicit, and what uncertainty the interface should absorb on the user’s behalf.</p><p>Design leadership starts earlier than most teams assume. Not when the UI is ready to polish, but when we ask what trust the experience needs to protect.</p><p>That is a decision-layer question. It cannot be answered by generating more screens. It requires someone to understand what the user is uncertain about, then design against that uncertainty before it becomes a product pattern.</p><h3>V. Designing With AI Requires Model Intuition</h3><p>AI literacy for designers is not primarily about knowing which tools to use. It is about understanding what kind of system we are collaborating with.</p><p>Large language models are powerful pattern systems. They can synthesize, compare, reframe, and generate plausible outputs from the context they are given. They do not carry our stakeholder context unless we define it. They do not know which trade-off matters in a specific organization unless we name it. They do not know when a confident answer is built on a weak assumption unless we know how to challenge it.</p><p>Prompting, in this sense, is not just asking for output. It is context architecture.</p><p>A strong designer does not only ask AI to “make this better.” A strong designer defines the audience, the constraints, the decision criteria, the risks, and the standard of quality before asking AI to generate options. Then, when the model responds with something polished, the designer does not accept coherence as truth.</p><p>In practice, model intuition means pausing when an AI-generated answer sounds convincing and asking a second layer of questions before treating the output as direction:</p><ul><li><strong>What assumption is this making?</strong></li><li><strong>What stakeholder context is missing?</strong></li><li><strong>What would make this recommendation wrong?</strong></li><li><strong>Which trade-off is being hidden by the smoothness of the language?</strong></li></ul><p>That pause is critical collaboration.</p><p>The future skill for designers is not simply using AI faster. It is knowing when to generate, when to compare, when to critique, when to distrust the model’s confidence, and when human accountability must take over entirely.</p><p>AI can help explore the solution space. Designers still have to frame the problem space.</p><h3>VI. The Judgment Era</h3><p><a href="https://web.mit.edu/curhan/www/docs/Articles/15341_Readings/Group_Performance/Edmondson%20Psychological%20safety.pdf">Amy Edmondson</a>’s research on psychological safety shows that team learning depends on whether people feel safe taking interpersonal risks, including asking questions, raising concerns, and admitting uncertainty.</p><p>This is not only a soft culture value. It is a precondition for good decision-making.</p><p>Designers are, or should be, architects of that condition. They are the people who notice when a team is converging too early on a solution before the problem is understood. They ask the question that reframes what everyone thought was already decided. They make the trade-off visible that others were hoping to avoid naming. They protect the user experience from the accumulated pressure of decisions made without the user in the room.</p><p>This is what I mean when I say the designer’s value is moving upstream. Execution still matters deeply. But in an AI-augmented environment, the highest-leverage contribution a designer can make is not always the artifact itself. It is the judgment that determines whether the artifact is solving the right problem.</p><p>My North Star is simple:</p><p><strong>I want to work in the moment before misalignment becomes momentum.</strong></p><p>That moment is not dramatic. It is a design review where someone asks, “What problem is this interface actually trying to solve?” and three people give different answers. It is a product brief that names the trade-off the team is making instead of hiding it inside scope language. It is a decision log that records not only what was chosen, but why, so that six months later, when the direction is challenged, the team has something more reliable to return to than memory.</p><p>These are small, structural acts. They do not always require formal authority. They require the habit of surfacing what is usually implicit: the assumption, the trade-off, the uncertainty, and the decision that has not yet been made clearly.</p><p>Designers are trained to see what users cannot articulate. In the AI era, the extension of that skill is learning to see what organizations cannot articulate: the decision that has not been made, the assumption that has not been named, the artifact that is moving faster than the team’s understanding of it.</p><p>That is the work.</p><p>It is not always glamorous. It does not always produce a new deliverable. But it determines whether the deliverables that do get produced are solving the right problem.</p><p>We are not here only to build screens.</p><p>We are here to build the logic that makes the screens matter.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=db8bd9912104" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How a UX Designer Approaches Etsy: Turning a Shop into an Experience]]></title>
            <link>https://medium.com/artcenter-graduate-interaction-design/im-how-a-ux-designer-approaches-etsy-turning-a-shop-into-an-experience-fee5a2292a6b?source=rss-99f231f2cf4c------2</link>
            <guid isPermaLink="false">https://medium.com/p/fee5a2292a6b</guid>
            <category><![CDATA[ecommerce]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[small-business]]></category>
            <dc:creator><![CDATA[Fei Yang]]></dc:creator>
            <pubDate>Mon, 27 Oct 2025 21:05:58 GMT</pubDate>
            <atom:updated>2025-10-27T23:42:43.993Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>When My Design Mind Met Etsy</strong></h3><p>As a UX designer, I’ve spent years thinking about how people interact with digital products. So when I opened my Etsy shop, I assumed that knowledge would make things easier, take great photos, write clear descriptions, and ship with care. Simple, right?</p><p>Then came my first <strong>one-star review</strong>. The buyer wasn’t upset about the product, they just didn’t understand how to care for it once it arrived. That moment hit me harder than I expected. But instead of taking it personally, I realized it was just another form of <strong>user feedback</strong>. That’s when it clicked: I wasn’t just selling cactus pads, I was designing a <strong>user experience</strong>. And like any product, it needed research, iteration, and empathy.</p><h3>Mapping the Customer Journey</h3><p>I began mapping the full <strong>customer journey</strong>, from discovering my shop to browsing, purchasing, unboxing, and caring for the product. Each step was a design opportunity.</p><p>At first, everything worked fine. My listings had clean photos and optimized SEO. But something felt off, people liked the products, yet few left reviews or returned. I started asking myself: <strong><em>What’s missing between “add to cart” and lasting impression?</em></strong></p><p>So, I redesigned the first touchpoint: <strong>information clarity.</strong> My original care guide was text-heavy and intimidating. I turned it into a <strong>visual step-by-step card</strong>, complete with warm icons and numbered actions. The difference was immediate, reviews started mentioning how “easy” and “thoughtful” the experience felt.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GRkSHAzqP4UKb7ERv9d6Uw.jpeg" /><figcaption><em>From text-heavy to visual storytelling.</em> I redesigned my plant care guide using illustrated steps instead of long paragraphs. The result? Fewer customer questions and more positive reviews about clarity and design.</figcaption></figure><h3><strong>Designing Emotion into Packaging</strong></h3><p>In UX, small moments build trust. The same applies to packaging. Originally, I mailed out my cactus pads in plain boxes with minimal branding. It worked, but the design didn’t <em>feel</em> special. So I developed the unboxing experience using desert-tone wrapping, recycled paper, and a handwritten thank-you note. I even added a small bonus cutting as a surprise. Soon, customers began sharing photos of their packages online, not because I asked them to, but because it sparked joy. That’s when I learned that the product ends when the user smiles, not when it ships.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4WIt1Dms_DmGCcQfcFjQBg.jpeg" /><figcaption><em>Packaging is the first touchpoint of emotion.</em> I moved from a plain wrap to a branded unboxing moment, eco paper, thank-you cards, and a small free gift. It transformed “a purchase” into a memory customers shared online.</figcaption></figure><h3>Data-Driven Decisions and Iteration</h3><p>I also started validating design decisions through small <strong>A/B tests</strong>, just like in product design.</p><p><strong>Visual Presentation:<br></strong>I tested two main cover photo styles: a clean white background versus a lifestyle image with texture and warmth. Tracking clicks and favorites showed that lifestyle photos drove far more engagement.</p><p><strong>Pricing Strategy:<br></strong>Next, I compared “product price + shipping” versus “free shipping with a slightly higher total.” To my surprise, even though Etsy rewards free-shipping listings in search, customers prefer paying shipping separately. They found it more transparent and trustworthy.</p><p>These experiments showed me that <strong>perception often outweighs platform logic</strong>, visual hierarchy and pricing psychology affect decisions the same way UI patterns do in apps.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oSiGZwX8tQdOgtAgTZXXNA.jpeg" /><figcaption><em>Visual consistency builds trust.</em> I ran an A/B test between two product thumbnails: one styled photo vs. one natural shot. The styled image (A) outperformed the raw version (B), proving that presentation directly impacts perceived value.</figcaption></figure><h3><strong>Closing the Loop: Beyond the Purchase</strong></h3><p>After improving visuals and packaging, I wanted to keep the connection alive post-purchase. So I added a <strong>QR code</strong> that linked to my plant-care tutorials and TikTok page. This small detail turned one-time buyers into returning customers who shared their plant growth stories months later.</p><p>That’s when I realized my Etsy shop wasn’t just a store, it had become a <strong>community touchpoint</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LiAnH5aqlmaR8OgKcjCHsA.jpeg" /><figcaption><em>Closing the loop from offline to online.</em> Each plant now ships with a QR code that leads to care videos and behind-the-scenes clips. This simple addition extended the customer journey beyond delivery, turning a product into an experience.</figcaption></figure><h3><strong>What I Learned</strong></h3><p>Running an Etsy shop through a UX lens taught me that design isn’t just for screens, it’s for experiences.</p><ul><li><strong>Good <em>visuals</em> attract, but good <em>experiences</em> retain.</strong></li><li><strong>Every touchpoint is a story opportunity.</strong></li><li><strong>Small tests lead to meaningful insights.</strong></li><li><strong>Empathy builds trust: in apps, products, or packaging.</strong></li></ul><p>Once I started treating my Etsy shop like a product, everything changed. Conversion rates improved, reviews felt genuine, and customers returned, not just for the plants, but for the care behind them. More importantly, it reminded me that design is never about selling more things. It’s about creating moments people remember.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fee5a2292a6b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/artcenter-graduate-interaction-design/im-how-a-ux-designer-approaches-etsy-turning-a-shop-into-an-experience-fee5a2292a6b">How a UX Designer Approaches Etsy: Turning a Shop into an Experience</a> was originally published in <a href="https://medium.com/artcenter-graduate-interaction-design">ArtCenter Graduate Interaction Design</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>