Ethnography in UX means observing everyday behavior and asking why, not just asking users what they want — as shown in a gift-giving project where younger adults personalized gifts (identity) while older adults preserved traditions (responsibility). The lesson: products exist inside social relationships (Venmo emojis, Spotify playlists), and observing workarounds can reveal features — like Bank of America's "Keep the Change" (round-up savings), which came from watching mothers round up checkbook entries, not user requests
What a UX audit is actually testing (and why most designers get it wrong)
A UX audit that starts with a checklist is a "presence check," not a real audit — it tells you whether elements exist, not whether they function (a returns policy in legal language, a ghost "Add" button, cross-sells before trust signals). Key tests: the 5-second test (does navigation require translation?), the scroll test (price, rating, CTA, trust signal visible before scrolling), the thumb-only test (tap targets on mobile), and edge cases — then prioritize by funnel stage
NNG: Does Your Form Really Need a Dropdown List?
An NN/g guide on dropdown lists: they work best in a narrow sweet spot (5–10 options, secondary to the main task, or part of a grouped layout) — avoid for too few options (radio buttons), too many (combobox), familiar data (text input), or visual comparison (button grids). Dropdowns are a tradeoff, not a default: ask how many options, whether users need to see them all, and whether the layout benefits from hiding them
AI: Agentic AI is the perfect machine for creating unused documents faster
Agentic AI is the perfect machine for creating unused documents faster — it produces polished artifacts that look professional but contain no real insight, because most organizations are built to receive familiar forms, not to think. The danger isn't bad work looking bad; it's mediocre work looking better than ever, and the real test is whether the artifact changes a decision, not whether it fills a template
Prototyping: The Neuroscience of UX Design - A Complete Guide
A guide connecting neuroscience to UX: the 50-millisecond verdict (visual appeal judged before conscious thought), Fitts's Law (target size/distance), Hick's Law (more options = slower decisions), and working memory limits (~3-4 items) explain why best practices work. Good UX strips unnecessary cognitive load; as AI generates interfaces faster, understanding these mechanisms may be the human designer's last edge
Experience: Research is not a deliverable
A case study on building qualitative research in a metrics-obsessed organization: the author trained customer support agents (trained to give answers, not ask questions) to conduct semi-structured interviews — turning problem-solvers into empathetic listeners through mock interviews (scores went from 6s to 10s). The result was a self-sustaining research machine with an AI analysis pipeline — proving research is not a one-off deliverable, but a discipline built into the organization's culture and workflow
Design: Rethinking Sparkles in the Age of AI
HP's UX research found sparkle icons now signal "AI" to users regardless of style — subtle differences went unnoticed (64% saw no difference) and didn't convey "AI-ness." Context matters: sparkles work on familiar AI territory (chat, images) but confuse on printers/documents; reserve them for real AI actions and add context in unfamiliar places
Basics: The Difference Between What Customers Say and What They Mean
A short reflection on a core product insight: customers are excellent at describing their problems but not always good at describing solutions — teams mistake requests (more filters, export button) for needs, build what was asked, and still leave the original problem unsolved. The job is to uncover intent behind requests: ask "what are customers trying to accomplish?" not "what do they want?" — because the insight comes from understanding why the request exists
@uxdigest
