UX Digest - лучшие посты

Nine Design Fixes for Your UX Research Reports Nine practical ways to make research reports easier to read and harder to ignore: clearer structure, stronger hierarchy, better contrast, more whitespace, consistent charts, and reusable templates. The authors make a very UX-ish point — a research report is itself an interface with two users: the researcher creating it and the stakeholder reading it. MeasuringU even usability-tested its own report template and went through four iterations before settling on the current version The user belongs to everyone, which is the same as belonging to no one When everyone on a product team “owns the user”, responsibility for actually understanding them can disappear. The author shows how overlapping research by UX and Product can turn into a data-quality problem when interview framing and methods aren’t shared. His practical fix is surprisingly small: agree upfront who researches, who is consulted, and who makes the call NNG: Using AI for UX Work - Study Guide NN/g’s practical map of where AI already fits into UX work — from research planning and analysis to prototyping, writing, service design, and workshops. The guide is less about “AI will replace UX” and more about choosing specific tasks where it can genuinely help without outsourcing judgment. The research section alone covers synthetic users, AI-moderated interviews, survey writing, digital twins, and why analysis still shouldn’t be handed over completely to AI Experience: Learnings from Experience Design Micro-Projects at NID A two-week experiment in designing tiny experiences around emotions like restlessness, adventure, and optimism — often with just a day to research, prototype, and test. The projects range from an endless staircase timer to chocolate as a trigger for hope and motion-activated “fireflies”. The best lesson comes at the end: even a handful of people was enough to break assumptions and reshape almost every prototype AI: The feature is cheap to code, so we might as well build it. (Probably not) AI makes features dramatically cheaper to build, but implementation is only a fraction of their real cost. Every new feature adds maintenance, support, friction, complexity, and eventually makes removing it a product decision of its own. The uncomfortable takeaway: when almost anything can be built, deciding what not to build becomes the harder skill Opinion: What first dates can teach us about becoming better, safer researchers What first dates can teach researchers about rapport, awkward silences, knowing when to end a conversation, and trusting your instincts. The authors connect familiar dating situations with practical research skills — from making interviews feel less like a checklist to planning fieldwork with exits, check-ins, and even a code word. A useful reminder that researcher safety deserves as much preparation as participant safety Basics: Persona Is Not a Document, It’s a Decision Tool A persona is useful only when it helps a team make product decisions, not when it simply describes age, job title, and favourite apps. The author suggests building personas around context, goals, pain points, behaviour, motivation, and decision drivers — then using them to prioritise features and UI choices. A good test is simple: can your persona help explain what you should avoid building? @uxdigest

Unlocking UX Secrets: How 1 Hour of Session Recordings Transformed Trendyol Seller Panel Just one hour of watching session recordings (Hotjar) revealed a massive pain point: sellers suffering the "scroll of death" for bulk actions, forced to scroll back to the top after selecting items. Three iterative designs later, a bottom-positioned sticky toolbar triggered by checkbox selection won—proving that quick, unbiased observation in natural workflow can uncover insights no survey or interview could, without spending extra budget or bothering busy users NNG: The Custodial Era of UX - Cleaning Up After AI AI lets teams build faster than UX can evaluate, creating UX debt: interfaces that "technically function" but confuse users, overload features, and skip foundational questions about whether something should exist at all. UX's new custodial role isn't just janitorial cleanup—it's triaging what to keep, accelerating evaluation (rapid user panels, checklists), and adding UX knowledge to generation (Design.md, design systems)—because the goal is to prevent the next mess, not just clean it up Prototyping: Rethinking Data Visualisation - A UX Approach To Dashboards That Actually Drives Decisions Most dashboards fail because they're built from available data, not from a specific operational question—so they show numbers without driving decisions. The fix is upstream: define context (what question are we answering?), audience (familiarity + accountability), and insight (what should change after seeing this?), then design the visualization to serve that, not the other way around—because the right chart isn't about simplicity, it's about appropriate complexity for the person and the decision at hand AI: Thoughtful AI implementation for UXR leaders AI skeptic or not, set a north star: AI should support, not replace, research quality (craft, assessment, revision). Preserve core skills with clear guidelines (don't use AI for research questions, do use it for data cleaning, label AI output), and frame every tool conversation around risk vs. reward—because in the rush for efficiency, we risk losing the human processing time that produces real insights Design: My “Book of Sorcery” Taught Me Design Before the Internet Did Before the internet, a fine art diploma taught design through observation and limitation: hand-drawn ads, lettering from a book of 30 fonts, and a physical "book of sorcery" of clippings from magazines and packaging. That instinct to collect and trust what "looks right" later met research training—behavioral psychology, HCI, accessibility—transforming instinct from the answer into the first question: why does it look right, to whom, and under what assumptions? @uxdigest

The gap between what user research surfaces and what gets built It isn't a quality problem—it's a translation problem: qualitative insights don't become backlog items because there's no formal process to bridge the formats. The fix is to embed decision-makers in synthesis, translate findings into prioritization-ready format (impact + effort), and track their status like bugs—otherwise, research remains an expensive activity with no measurable value NNG: AI Can't Replace Real Research in Empathy Mapping AI can help organize existing research into empathy maps, but it cannot replace real user evidence—because plausible, generic quotes (e.g., "I wish the app would ask before swapping") are not the same as specific, messy details from real sessions (e.g., "It swapped my oat milk for whole milk and charged me before I saw the notification"). Before reaching for AI, ask: am I organizing real data or generating data I don't have? If the answer is the latter, the map needs more research, not more AI Prototyping: Using content design to fix UK emergency alerts The UK emergency alert about wildfires was well-intentioned but failed its users: dense text, redundant language, weak visual hierarchy, and unhelpful signposting made it harder to scan in a moment of stress. A content redesign—shorter sentences, direct "do not" commands, clear paragraphs, and a specific URL—cut reading time from 19 to 14 seconds and dropped the reading age from 17.6 to 14.6, proving that in an emergency, every word and second counts Experience: We need a survey — a UX research story When a stakeholder says "we need a survey by Tuesday," the real skill isn't asking for more time—it's starting with the question, not the method, and doing credible research within imperfect conditions. Fast research means reducing scope, focusing on the most critical questions, and making deliberate trade-offs (e.g., 5 relevant users instead of 20 later)—and sometimes the method itself evolves as findings emerge, because research is about following the evidence, not sticking to a plan AI: Craft still matters, but it’s about outcomes Craft isn't about pixels anymore—when AI makes production cheap, real craft moves to choosing the right problem and curating AI output with taste. To keep it, write your standards down (so the machine applies your rules), keep research a weekly habit, and always put a human review at the end—because AI fails quietly, and fluent confidence without verification is just fiction Design: Why People Ignore Perfectly Designed Buttons People ignore perfectly designed buttons not because of color, size, or placement, but because of the trust problem (uncertainty about what happens next), decision fatigue (too many choices already made), and identity questions (not ready to be the person who clicks it). The fix isn't polishing the button—it's building the right emotional state, trust, and narrative momentum before it appears, because a button is just the answer to a question the experience needs to ask first @uxdigest

Five Ways to Use Rapport When Moderating Good moderation isn't just reading a script—it's building trust through preparation (know your audience's lingo), deliberate ice-breaking, and a verbal toolkit (purposeful pauses, neutral expressions, camera eye contact). But watch for the failure mode: mutual people-pleasing that makes disagreement feel impolite, and use rapport as a final screening step—because a few warm-up questions can catch a participant who never should have qualified Why Users Don’t Experience Your Interface the Way You Designed It The gap between your Figma file and reality is shaped by the curse of knowledge (you can't unsee your design decisions), context collapse (users are distracted, stressed, rushing), mental model mismatches (they expect different logic), and accessibility oversights. The fix isn't more testing, but treating it as genuine inquiry: test in real contexts, involve people who disagree with you, and cultivate humility—because users experience your design through their own reality, not yours NNG: We Still Need Long-Form Copy Despite shrinking attention spans, long-form copy still matters—because some things are inherently complex, require trust, or need SEO and AEO benefits. The key isn't cutting words, but making them scannable with summaries, bullet points, callouts, bolding, and visuals to improve comprehension and engagement with content over 1,000 words Prototyping: Designing High-Performance User Experiences for Data-Heavy Dashboards and Analytics Products Consumer design patterns (clean cards, progressive disclosure) fail in data-heavy expert tools—experts need density, bulk actions, and state preservation, not simplicity. The craft lies in balancing density, clarity, and speed: dense tables with strict alignment, keyboard shortcuts, and transparent permissions—because friction isn't clicks, but the gap between how work actually happens and how software forces it AI: Researcher-in-the-loop Instead of asking if AI will replace researchers, ask who keeps research honest when everyone can self-serve. The researcher-in-the-loop model makes researchers governors of the system: they set rules (sourcing, confidence, escalation), route low-risk questions to AI tools, and keep high-risk ones for humans—because a confidently wrong AI is worse than no answer Design: How Good Design Will Save the World Good design isn't loud—it's invisible, seamless, and accessible, solving problems without being noticed. The best designers focus not on articulating every use case, but on delivering what people really want (like the snooze button: a simple reprieve in a complex world), and make accessibility the core, not an afterthought—because true design considers both the space and the user, creating experiences that work for everyone, not just the able-bodied @uxdigest

Designing Flow in SSX 3: A Cognitive Psychology & Game UX Analysis A cognitive psychology analysis of SSX 3 shows how the game sustains flow through a risk/reward trick system (adrenaline → Ubertricks → Super Uber), adaptive audio that gives real-time feedback, and a progressive mountain structure that balances challenge with skill. It satisfies Self-Determination Theory (competence, autonomy, relatedness) through trial-and-error learning, free exploration, and community-building — showing how arcade design can coordinate mechanics, audio, and progression for deep immersion Atomic UX research: How to pilot a repository with a Claude skill A Claude skill automates atomic UX research in Notion — handling experiments, facts, insights, and recommendations while cross-referencing new findings against existing data to surface patterns. It saves ~4 hours per study by cutting administrative work, keeping humans in the loop for decisions like interpreting nuance and framing questions 🎥 NNG: 3 Steps to Scope Your Broad Research Study Proper scoping is tied to timelines, research efficiency and the usefulness of your outputs. Learn how to scope your studies for optimal impact Prototyping: Why a Single Red Dot Can Hijack User Behavior for Weeks A red dot isn't just a UI element—it's a psychological trigger that creates an open loop in the brain, hijacking attention through the Zeigarnik effect and triggering a cortisol spike via evolutionary color-coding. When combined with variable rewards, this tiny circle can turn a productivity tool into a slot machine, creating compounding micro-anxieties that degrade focus and trust over weeks, not minutes AI: Measuring the UX of AI Measuring the UX of AI chatbots starts with standard UX metrics (usability, usefulness, effectiveness), but to truly understand the experience, you need to dig into specialized constructs like AI Productivity, Trust, Dependency, Anxiety, and Personification—each requiring validated items. The authors propose a framework with initial item sets for these five dimensions, but note that the real work lies in psychometric validation with real user data to see which items actually hold up and how they connect to higher-level outcomes like brand attitude and continued use Opinion: From UX to AX - Why I Think We’re Not Just Redesigning Interfaces, We’re Redesigning Trust We're moving from UX to AX (Agentic Experience), where the core challenge isn't mapping deterministic flows, but designing for probabilistic systems that act autonomously—shifting the design job from screens to behavior and relationships. The five real pillars of this new discipline aren't about capability, but about restraint: intent alignment, controllability, explainability, emotional intelligence, and co-evolution, because trust isn't earned by how much an agent can do, but by knowing exactly when to pause and ask The channel is going on vacation until the end of the summer, thank you for being with me all this time, see you in September @uxdigest

The Psychology of Why Confirmation Dialogues Get Ignored Confirmation dialogs are a psychological failure—they force a slow, analytical System 2 response onto a fast, automatic System 1 workflow, meaning users click through them without reading, driven by habit and the desire to resolve an interruption. The fix isn't better copy or redder buttons, but designing for reversibility (undo, soft deletes) and using 'good friction' like typing-to-confirm for truly irreversible actions, because a warning that appears daily becomes invisible noise NNG: Why You Need Mentorship and How to Get It Right Mentorship isn't just for juniors—research shows it leads to more promotions, higher pay, and less burnout for both mentees and mentors, yet many avoid it due to misconceptions about its value or who it's for. The key to success lies in choosing the right mentor (internal or external, candid and compassionate), setting clear goals and cadence, and actively putting advice into practice—because a mentor who sees you act on their guidance is far more invested in your growth Prototyping: The Beautiful Lie of the Progress Bar - Why We Trust Fake Progress Over Honest Waiting Fake progress bars work because uncertain waits (spinning wheels) create anxiety, while progress bars convert uncertainty into a finite wait, occupying attention and managing the Zeigarnik tension — users prefer the lie for the sense of forward momentum. Use the Three Tiers framework: indeterminate (30s or step-based, honest step counts); never compress step-based workflows — it breaks trust and triggers reactance Experience: How I designed a system that 95% of non-technical users could use without support The secret to 95% of non-technical users using a complex system without support isn't removing features, but matching the software to their existing mental models—like organizing a poultry platform around "batches" instead of individual birds, because that's how farmers actually think. The real breakthrough came from observing their workarounds (physical calendars, crate-based counting) and asking open-ended questions, then designing workflows that felt familiar enough that they barely had to learn them at all AI: I Stopped Waiting for Engineers to Validate My Ideas The real bottleneck for designers isn't ideas, but waiting for engineering validation—so with AI-assisted development, you can now prototype working solutions yourself, turning problems into interactive experiments in days instead of weeks. This shift isn't about replacing engineers, but about expanding design's role from creating screens to testing systems and decisions, proving that the hardest part of a transit app isn't the map, but reducing the uncertainty of waiting Front: Why a Website Works for Some Visitors, but Not Others A website isn't a fixed object, but a partnership between a server and a browser—and every visitor brings a unique environment (cache, cookies, extensions, device) that can make the same site work flawlessly for one person and break for another. The key isn't asking "does it work?", but understanding where the problem lives: try an incognito window to isolate browser-specific issues, and for owners, monitor analytics for early warnings before a single visitor complains Opinion: The Invisible Weight - Why Your “Simple” Interface is Exhausting Your Users Minimalism isn't about empty screens, but about invisible cognitive taxes—and aggressively stripping away visual cues often forces the user's brain to work harder, not less. A clean interface can actually be cognitively exhausting when it hides context behind ambiguous copy or progressive disclosure, turning 'simple' into a frustrating puzzle for your users @uxdigest

NNG: No New Name Has Replaced “UX” Despite ongoing debates about renaming the field, a survey of 604 professionals found that "UX" remains the dominant, spontaneous term (appearing in 70% of responses), while alternatives like "experience design" or "human-centered design" are fragmented and rare. The data suggests that job roles shape terminology (e.g., product managers say "product"), but no single alternative has emerged as a replacement—so the real work isn't finding a new label, but clearly communicating UX's contribution to decisions, outcomes, and risk reduction Prototyping: Why Skeleton Screens Feel Faster (Even When They Aren’t) Skeleton screens aren't just a visual trick—they work because they replace the anxiety of an unknown wait with a predictable structural map, letting the brain relax and perceive time differently. But the moment that skeleton doesn't perfectly match the final layout, you trigger a devastating layout shift that shatters user trust and makes the product feel broken, proving that the illusion of speed is only as good as its stability AI: Designing with web standards - The playbook for this AI moment We're living through 1999 again—AI interfaces (ChatGPT, Claude, Gemini) are the new browsers, each with its own proprietary patterns, and without shared standards, we're rebuilding the same broken, inconsistent experiences across every tool. The playbook from Jeffrey Zeldman's web standards movement is clear: name the patterns (confidence, citations, reasoning) before vendors lock them in, build a coalition of practitioners, and make the business case (cost, reach, accessibility)—because standards win when agreeing becomes cheaper than diverging Case Study: How We Redesigned TikTok Like a Real Product Team in Five Weeks Through 3 rounds of testing and iteration, the team used Figma Make to prototype solutions for button control, algorithm transparency, and viewing time — increasing brand trust from 5.5 to 7.5 (+36%). Key changes: swipe-to-dismiss "Not Interested," algorithm dashboard with feed tuner, and Screen Time Pledge (forced exit). 81.8% said they'd recommend the redesigned version Experience: We Thought We Understood the User A product manager thought they understood the warehouse process from requirements docs — until visiting the site revealed workers had quietly moved a key step much earlier in the workflow, and nobody had filed a change request. The lesson: requirements capture the paper version, not real adaptations; discovery means going where the work actually happens and asking what the floor knows that you're missing Opinion: The Friction Paradox - Why Adding Steps to Your Product Actually Builds More Trust We've been taught that every click is a tax on the user, but in high-stakes moments, removing all friction doesn't build confidence—it triggers suspicion, making users feel rushed and unsure about their decisions. The secret isn't eliminating friction, but distinguishing bad friction (bureaucratic, confusing) from good friction (educational, confirmational)—where a deliberate pause or extra step acts like a speed bump, forcing the user to slow down and actually own their choice @uxdigest

Transitioning from Graphic Design to UX Moving from graphic design to UX doesn’t mean starting over. Most core skills are directly transferable and serve as a strong foundation to become a great UX designer Prototyping: The Enterprise Software Delusion - We Design for the Buyer and Blame the User Enterprise software fails because buyers (executives) buy for compliance, but users get bloated interfaces designed for audits, not action — "resistance" is really just increased cognitive load. The fix: measure Time to Core Value, involve users in demos, reduce mandatory fields, and absorb complexity to present a simple path — align incentives with human reality AI: I Didn’t Start With AI, I Started With People A UX researcher studied top human agents to inform an AI assistant: the best agents don't just answer — they first understand the customer's situation, make sure solutions are understood, and introduce products by asking about habits first, then recommending benefits (not promotions). The key insight: AI should start by understanding context, not answering; recommendations should feel like helpful suggestions, and some situations still need human judgment — AI should prepare and escalate, not replace Case Study: Improving Recommendation Controls on Netflix Users separate rating (thumbs up/down for watched content) from dismissing unwanted recommendations, and they're unwilling to spend time managing preferences — they need a lightweight, visible action on the card. The solution: a context menu with optional reasons (already watched, not interested, fewer like this) followed by confirmation with Undo — because control alone isn't enough; the interaction must feel worth the effort Opinion: Everything you know about UX is probably wrong A reflection on "User Experience Foundations": UX isn't a deliverable or a role — according to ISO 9241-210, it's a person's perceptions and responses, meaning you can only design the conditions for experience, not the experience itself. Key principles: you are not the user, behavior-based research (contextual inquiry) beats opinion-based surveys, users must be involved throughout, and UX accounts for anticipatory, episodic, and cumulative experience — not just the moment of use Basics: Product Managers Don’t Need More Data. They Need Better Questions Product managers don't need more data — they need better questions; teams with 14 dashboards can measure everything but still can't answer "should we keep building this" because data tells you what happened, not what to do next. The fix: before collecting any metric, write down the exact decision it's meant to inform and what answer would change it — good questions must be built deliberately, before instrumentation Interesting: Understanding shisha as a different experience of smoking NHS research found that shisha smokers often don't see it as "proper smoking" — it's social, seasonal, and positive, so they don't report it in standard smoking questions designed for regular cigarette use. Testing "Have you smoked any of these tobacco types for 1 year or longer?" caused confusion for episodic holiday users, showing that shisha requires different mental models and question design, not just being slotted into cigarette-based journeys @uxdigest

Agile UX and its associated challenges Don Norman's critique of Agile UX: "Norman's Law" — on the day a project is announced, it's already late and over budget — because Agile sacrifices user research for coding, when research should be a continuous practice, not a phase. The fix: continuous research, Design Sprints to compress understanding before coding, and reframing procrastination as "time to think" — building feature by feature without overall coherence leads to products that fail NNG: UX-Context Design - Using UX Knowledge to Inform AI-Generated Design NN/g introduces "UX-context design" — as AI generates more interfaces, research output shifts from human deliverables to machine-readable context (standards, research insights, user/world models) that guides AI tools and prevents generic output. Examples: DESIGN.md (visual identity) and UX.md (research, interaction standards, glossary) — never-finished files that live alongside code and make research a continuously curated source of truth, not a one-time handoff AI: AI in UX Research - What Changes When the Researcher Is Also the Model A critical framework for AI in UX research: distinguish three roles — assistant (transcribing real human data, low risk), proxy (synthetic personas, treat as hypotheses, not findings), and researcher (agentic AI running the whole pipeline, highest risk, especially when the same model generates and interprets data). The key principle: no model has a nervous system — keep a human in the judgment call, check for circularity, and treat synthetic output as a hypothesis, not a finding Opinion: Beyond Jobs to Be Done - Why Technical Fluency Is Now a Research Requirement JTBD works from individual intention, but in technical categories (enterprise software, AI, infrastructure), outcomes are shaped by non-human actors (data rules, integrations, models) — treating these as background produces research that's right about what buyers want but wrong about what happens. The fix: actor-network theory — map all actors (human and non-human) that change the outcome, requiring researchers who can read technical arrangements, not just interview buyers Basics: Basic Tools & Techniques Every Product Manager Should Master for User Research A product manager's guide to user research: surveys for quantitative feedback at scale, interviews to uncover the "why," and observation tools (Google Analytics, Hotjar, Amplitude) to watch what users actually do — because people often say one thing and do another. The cycle: survey → interview → observe → build → measure → learn; combine qualitative and quantitative data to validate assumptions before making decisions @uxdigest