Claude Prompts for UX/UI Design Work

Picture this: a solo product designer at a 20-person startup just wrapped five usability test sessions, has 90 minutes of raw transcript per session, and needs a findings summary for a stakeholder meeting in two hours. This is precisely the kind of pressure where claude prompts for ux design either save the day or produce a summary too generic to act on.

The Problem This Designer Faced

She had five hours of transcript, no research ops support, and a stakeholder audience that wanted specific, actionable findings — not a vague "users found the flow confusing" line they'd heard in every previous research readout.

The Wrong Approach

Her first attempt pasted a full transcript and asked:

Summarize the usability issues in this transcript.

The output was accurate but shallow — a list of individually true observations with no sense of which ones mattered most, or which ones the design could realistically address before the next release.

The Correct Prompt

Here is a usability test transcript for [feature being tested]:
[paste transcript]

Identify:
1. Every point where the participant hesitated, backtracked, or
   expressed confusion — quote the moment briefly (paraphrase, don't
   quote verbatim) and note the specific UI element involved
2. Rank these by apparent severity: did it block task completion,
   slow it down, or just cause a comment?
3. For the top 3 highest-severity issues, suggest what specifically
   about the interface likely caused the confusion — not a generic
   "improve clarity" fix

What this does: Anchoring each issue to a specific UI element and a severity ranking is what turns a list of observations into something a design team can prioritize. The "not a generic improve clarity fix" instruction pushes Claude toward a more specific, actionable hypothesis about the actual cause.

Results and What Changed

Running this prompt across all five transcripts surfaced a pattern none of the individual sessions made obvious on their own: three of five participants hesitated at the exact same step, a confirmation dialog with ambiguous button labels. That single finding became the headline of her stakeholder readout, and the fix shipped within two weeks — a much faster turnaround than the vaguer "users found onboarding confusing" findings from a previous round had produced.

⚡ Pro tip: After summarizing each transcript individually, ask Claude to compare all five summaries and identify the issue that recurs most often. Patterns across sessions are the strongest evidence in a research readout — much stronger than any single session's most dramatic moment.

⚠️ Common mistake: Summarizing transcripts one at a time and stopping there, without a cross-session synthesis step. The most persuasive finding in usability research is almost always "this happened to multiple people," and that only surfaces if you explicitly ask for the comparison.

How to Apply This to Your Situation

For synthesizing open-ended survey responses instead of interview transcripts:

Here are 150 open-text survey responses to "what's frustrating about
[feature]": [paste responses]

Cluster these into 5-7 themes. For each theme, note approximately
what percentage of responses fall into it and paraphrase one
representative comment per theme — no verbatim quotes.

What this does: This adapts the same severity-and-pattern logic to a much larger, unstructured dataset — the kind of survey data that usually sits unread in a spreadsheet because nobody has time to manually code 150 responses.

Scenario: UX researcher at a healthcare startup analyzing accessibility feedback A UX researcher used the transcript-analysis prompt on accessibility-focused usability sessions with screen reader users, specifically asking Claude to flag any moment where the screen reader announced something inconsistent with what was visually on screen — a class of bug easy to miss without that framing.

Scenario: Product manager triaging feature request feedback A PM without formal UX research training used the survey-clustering prompt to make sense of a backlog of 200 unstructured feature requests submitted through a support form, turning an unreadable spreadsheet into seven clear themes she could bring to a roadmap discussion.

Scenario: Design lead building a case for a redesign A design lead used cross-session synthesis findings, built the way described above, as the evidentiary backbone of a pitch to leadership for a checkout redesign — leading with the pattern across five sessions rather than her own opinion carried more weight in that conversation.

Next Steps

The core lesson is the same across every claude prompts for ux design use case: ground the analysis in specific UI elements and look explicitly for patterns across sessions, not just within one. Once you've got a transcript-analysis prompt that reliably surfaces this kind of finding, save it in PromptABCD so your next round of usability testing starts with a proven synthesis method instead of a blank prompt and a ticking clock.