30+ AI Prompts for UX Design: Research, Wireframes & Testing

Why UX Designers Need Structured AI Prompts

Unstructured UX prompts produce generic outputs 72% of the time (AI Prompt Architect platform data, Q2 2026). When designers apply the STCO framework — Situation, Task, Context, Output — that figure drops to 19%. The difference isn't marginal; it's the difference between receiving a boilerplate persona and one that includes behavioural triggers, accessibility needs, and journey-stage context.

The STCO framework maps naturally onto UX workflows:

Only 11% of UX prompts in our dataset included accessibility constraints such as WCAG references or screen-reader context. Those that did produced outputs 2.8x more likely to include inclusive design considerations (AI Prompt Architect benchmark, n=5,247). This finding drove us to embed WCAG callouts throughout the templates below.

A necessary caveat: AI cannot replace user empathy. These prompts accelerate synthesis, not judgement. Every AI-generated deliverable should be validated against real user data and reviewed by a qualified designer before entering production workflows. For broader guidance on structuring prompts across disciplines, see our prompt engineering best practices guide.

User Research Prompts: 8 Templates

Our Prompt Scorer evaluated 1,340 user-research prompts and found that prompts specifying a research method scored 28% higher on output specificity (AI Prompt Architect platform data, Q2 2026). Adding demographic segmentation parameters increases output actionability by 31%. The templates below encode both principles.

Always validate AI-synthesised research against raw participant data. AI can structure interview guides and synthesise themes, but it cannot observe a participant's body language or detect sarcasm in a diary entry.

1. Interview Script Generator

Situation: I am conducting [number] semi-structured interviews for [product/feature] targeting [user segment].
Task: Generate a 45-minute interview script covering [research objectives].
Context: Participants are [demographics]. Prior research indicates [known pain points]. We follow [ethical guidelines, e.g. BPS or APA].
Output: Script with warm-up (5 min), core questions (30 min), and wrap-up (10 min). Include probing follow-ups for each core question. Format as a numbered list with timing annotations.

2. Survey Question Crafting

Situation: We need a [quantitative/qualitative/mixed] survey for [product] distributed to [n] users via [channel].
Task: Write [number] survey questions measuring [variables, e.g. task satisfaction, feature discoverability].
Context: Target completion time is [X] minutes. Use [Likert scale / multiple choice / open-ended] format. Avoid leading questions and double-barrelled items.
Output: Questions in table format with columns: Question Number, Question Text, Response Type, and Rationale.

3. Diary Study Protocol

Situation: Running a [duration]-day diary study with [n] participants using [product/feature].
Task: Create a diary study protocol including entry prompts, submission schedule, and analysis framework.
Context: Participants are [demographics]. We are investigating [behaviours/emotions]. Entries will be submitted via [method].
Output: Protocol document with: daily prompts, example entries, escalation criteria, and a thematic analysis codebook template.

4. Contextual Inquiry Guide

Situation: Observing [n] users in their [environment, e.g. home office, retail floor] while they use [product/system].
Task: Generate a contextual inquiry observation guide.
Context: Focus areas: [workflow steps, pain points, workarounds]. Session length: [duration]. Recording method: [notes/video/audio].
Output: Checklist with observation categories, sample probing questions, and a post-session debrief template.

5. Affinity Mapping Synthesis

Situation: We have completed [n] user interviews for [project]. Raw notes total approximately [word count] words.
Task: Synthesise interview notes into affinity map clusters.
Context: Research questions were: [list]. Key themes emerging informally include [themes]. We need [number] top-level clusters.
Output: Hierarchical affinity map with top-level themes, sub-themes, and 2–3 representative quotes per sub-theme. Flag any contradictory findings.

6. Competitive UX Audit

Situation: Auditing [n] competitors in the [industry] space for [product].
Task: Produce a comparative UX audit across [specific flows, e.g. onboarding, checkout, search].
Context: Competitors: [list with URLs]. Evaluation criteria: [task completion, visual design, accessibility, mobile responsiveness]. Use Nielsen's 10 heuristics as the scoring framework.
Output: Comparison table with columns: Competitor, Heuristic, Score (1–5), Evidence, and Opportunity. Include a summary of top 5 differentiation opportunities.

7. Jobs-to-Be-Done Extraction

Situation: Analysing [n] customer interviews to extract JTBD statements for [product/feature].
Task: Extract Jobs-to-Be-Done statements in the standard format: "When [situation], I want to [motivation], so I can [expected outcome]."
Context: Interview transcripts cover [topics]. Target user segments: [list]. Prioritise functional jobs, then emotional jobs, then social jobs.
Output: Numbered JTBD statements grouped by job type. Include importance and satisfaction ratings where evidence supports them.

8. Accessibility Audit Brief

Situation: Conducting a WCAG 2.2 AA compliance audit for [product/URL].
Task: Generate an accessibility audit checklist covering all Level A and AA success criteria.
Context: The product uses [tech stack]. Known issues include [list]. User base includes [percentage] assistive technology users. Test with [screen readers, e.g. NVDA, VoiceOver].
Output: Checklist in table format: WCAG Criterion, Level, Pass/Fail, Evidence, and Recommended Fix. Group by POUR principles (Perceivable, Operable, Understandable, Robust).

Personas & Empathy Maps: 7 Templates

Persona prompts that include behavioural triggers produce outputs rated 'immediately usable' by designers 67% of the time, compared to 23% for basic demographic-only prompts (AI Prompt Architect platform data, Q2 2026). Similarly, empathy maps structured around the Says/Thinks/Does/Feels quadrants score 44% higher on completeness than free-form alternatives.

Proto-personas are starting points, not substitutes for real user research. Use these templates to scaffold hypotheses that you then validate through the research methods in the previous section.

1. Data-Driven Persona Synthesis

Situation: We have completed [n] interviews, [n] survey responses, and [analytics data] for [product].
Task: Synthesise a primary persona from the provided research data.
Context: Key behavioural patterns observed: [list]. Demographics: [ranges]. Goals and frustrations identified: [summary]. Include accessibility needs and technology proficiency.
Output: Persona card with: Name, Photo description, Demographics, Goals (3–5), Frustrations (3–5), Behavioural triggers, Preferred channels, Accessibility requirements, and a Representative quote.

2. Proto-Persona

Situation: Early-stage [product] development with limited user data.
Task: Create a proto-persona based on stakeholder assumptions and market research.
Context: Target market: [description]. Assumptions: [list]. Business goals: [list]. Flag all assumptions explicitly for later validation.
Output: Proto-persona card with clearly labelled "Assumed" and "Validated" fields. Include a validation plan listing which assumptions to test first.

3. Empathy Map from Transcripts

Situation: Processing [n] interview transcripts for [product/feature].
Task: Generate an empathy map using the Says/Thinks/Does/Feels framework.
Context: Research focus: [topic]. Participant segment: [description]. Include direct quotes in the "Says" quadrant.
Output: Four-quadrant empathy map with 4–6 data points per quadrant. Highlight contradictions between quadrants (e.g. says one thing, does another).

4. Journey-Stage Persona

Situation: Mapping persona behaviour across the [awareness/consideration/decision/retention] journey for [product].
Task: Create a persona variant for each journey stage showing how needs, emotions, and touchpoints evolve.
Context: Current journey map: [summary]. Key drop-off points: [list]. Channels used: [list].
Output: Table with columns: Journey Stage, Primary Goal, Emotional State, Key Questions, Touchpoints, and Pain Points.

5. Accessibility Persona

Situation: Designing [product] to be inclusive for users with [disability type, e.g. low vision, motor impairment, cognitive disability].
Task: Create an accessibility-focused persona representing users with [specific needs].
Context: Assistive technologies used: [list]. WCAG 2.2 level target: [AA/AAA]. Common barriers in current product: [list].
Output: Persona card including: Assistive technology setup, Daily usage patterns, Specific barriers encountered, Workarounds used, and Design requirements that would remove barriers.

6. B2B Buyer Persona

Situation: [B2B SaaS product] targeting [industry/role, e.g. operations managers in manufacturing].
Task: Create a B2B buyer persona that includes both individual user needs and organisational buying context.
Context: Buying committee roles: [list]. Budget authority: [level]. Decision timeline: [duration]. Integration requirements: [list].
Output: Persona card with: Role, Organisation size, KPIs they own, Buying triggers, Objections, Evaluation criteria, and Preferred content formats.

7. Persona Validation Checklist

Situation: We have [n] personas created for [product] and need to validate them against real data.
Task: Generate a validation checklist for each persona.
Context: Available data sources: [analytics, interviews, surveys, support tickets]. Personas were created [duration] ago. Product has [n] active users.
Output: Checklist with: Validation method, Data source, Sample size needed, Key metrics to verify, and Red flags that indicate persona revision is needed.

Information Architecture Prompts: 8 Templates

IA prompts that reference card-sorting data produce sitemaps 2.1x more structured than those without (AI Prompt Architect platform data, Q2 2026). STCO-structured IA prompts achieved 79% compliance with Nielsen Norman Group heuristics versus 41% for unstructured prompts — a gap that directly affects findability and task completion. For more on structuring prompts with the STCO method, see the STCO framework guide.

WCAG Note: All IA templates below include prompts for semantic HTML landmarks, skip-navigation links, and logical heading hierarchies. Screen-reader users rely on these structures to navigate efficiently.

1. Sitemap Generation

Situation: Redesigning the information architecture for [website/app] with [n] pages/screens.
Task: Generate a hierarchical sitemap limited to [n] levels of depth.
Context: Primary user tasks: [list]. Content types: [list]. SEO priority pages: [list]. Current analytics show [top pages, exit pages].
Output: Sitemap in indented list format showing parent-child relationships. Include page purpose annotations and suggested URL slugs.

2. Navigation Taxonomy

Situation: Designing primary and secondary navigation for [product] with [n] content categories.
Task: Create a navigation taxonomy that supports [n] user types and [n] key tasks.
Context: Card sort results: [summary]. Maximum nav items: [n] primary, [n] secondary. Mobile constraints: [hamburger/tab bar/bottom nav]. Accessibility: support keyboard navigation and ARIA landmarks.
Output: Navigation structure with labels, hierarchy, and mobile adaptation notes. Include ARIA role recommendations.

3. Content Audit Matrix

Situation: Auditing [n] pages of content for [website] ahead of a migration/redesign.
Task: Create a content audit matrix categorising each page by quality, relevance, and action needed.
Context: Business goals: [list]. Target audiences: [list]. Content age ranges from [oldest] to [newest]. Analytics available: [yes/no].
Output: Table with columns: URL, Page Title, Content Type, Quality Score (1–5), Traffic, Action (Keep/Revise/Merge/Remove), and Notes.

4. Card Sort Synthesis

Situation: Analysing results from a [open/closed/hybrid] card sort with [n] participants for [product].
Task: Synthesise card sort data into recommended content groupings.
Context: Cards: [n] items. Agreement threshold: [percentage]. Participant demographics: [summary].
Output: Recommended groupings with agreement percentages. Dendogram description. List ambiguous items that need further testing.

5. Tree Test Analysis

Situation: Evaluating a proposed IA through tree testing with [n] participants and [n] tasks.
Task: Analyse tree test results and recommend IA improvements.
Context: Success rate target: [percentage]. Current overall success: [percentage]. Problem tasks: [list with success rates].
Output: Analysis table with: Task, Success Rate, Directness, Time, First Click Accuracy, and Recommended Fix. Prioritise fixes by impact.

6. URL Structure Planning

Situation: Planning URL structure for [website] with [n] content types across [n] languages/regions.
Task: Design a URL structure that supports SEO, internationalisation, and user comprehension.
Context: CMS: [platform]. Current URL patterns: [examples]. Redirect requirements: [count]. SEO priority keywords: [list].
Output: URL structure template with examples for each content type. Include canonical URL strategy and hreflang recommendations.

7. Breadcrumb Logic

Situation: Implementing breadcrumb navigation for [website] with [n]-level hierarchy and [cross-linked content].
Task: Define breadcrumb logic including edge cases (multi-parent pages, filtered views, search results).
Context: IA depth: [n] levels. Cross-linked pages: [examples]. Schema.org markup required: [yes/no]. Accessibility: breadcrumbs must use nav landmark and aria-label.
Output: Decision table covering each page type with breadcrumb path, truncation rules, and schema markup examples.

8. Cross-Linking Strategy

Situation: Improving internal linking across [n] pages for [website] to support both UX and SEO.
Task: Design a cross-linking strategy that increases content discoverability and reduces bounce rate.
Context: Current average internal links per page: [n]. Orphan pages: [n]. Top landing pages: [list]. User journey priorities: [list].
Output: Linking matrix showing recommended links between content clusters. Include anchor text guidelines and a priority queue for implementation.

Wireframing & Layout Prompts: 8 Templates

Including viewport breakpoints in wireframing prompts produces responsive-ready outputs 3.4x more often (n=890, AI Prompt Architect platform data, Q2 2026). Prompts that reference a specific component library — such as Material Design 3 or a custom design system — achieve 52% higher developer-handoff accuracy. These findings align with broader patterns we document in our complete guide to AI prompting tools.

WCAG Note: Every layout template below includes prompts for focus order, touch target sizing (minimum 44×44 CSS pixels per WCAG 2.2), and colour contrast ratios. Responsive design and accessible design are not separate concerns.

1. Responsive Wireframe Specification

Situation: Creating wireframes for [page/feature] across [mobile 375px / tablet 768px / desktop 1440px] breakpoints.
Task: Specify wireframe layout including content hierarchy, component placement, and responsive behaviour.
Context: Design system: [name/version]. Key components: [list]. Content priority: [ordered list]. Accessibility: logical tab order, minimum 44×44px touch targets.
Output: Wireframe specification per breakpoint in structured text with: grid definition, component list with position, content priority annotations, and responsive adaptation notes.

2. Component Hierarchy

Situation: Defining the component hierarchy for [page type] in [design system].
Task: Create a component tree showing parent-child relationships and data flow.
Context: Atomic design methodology: [atoms/molecules/organisms]. Existing components: [list]. New components needed: [list]. State management approach: [description].
Output: Component tree in indented list format with: component name, type (atom/molecule/organism), props, and accessibility requirements (ARIA roles, keyboard interactions).

3. Layout Grid System

Situation: Establishing a layout grid for [product] supporting [n] breakpoints.
Task: Define a grid system with columns, gutters, margins, and baseline grid.
Context: Target devices: [list]. Content density: [low/medium/high]. Typography scale: [base size and ratio]. Design tool: [Figma/Sketch/Adobe XD].
Output: Grid specification table with: breakpoint, columns, column width, gutter, margin, and max-width. Include CSS Grid implementation notes.

4. Above-the-Fold Optimisation

Situation: Optimising above-the-fold content for [page type] on [device/viewport].
Task: Specify content hierarchy and component placement for the initial viewport.
Context: Primary CTA: [description]. Key metrics: [conversion rate, bounce rate]. Current LCP: [time]. Image/video requirements: [specifications].
Output: Prioritised content list with estimated pixel heights per element. Include loading strategy (eager vs lazy) and performance budget.

5. Mobile Navigation Pattern

Situation: Designing mobile navigation for [app/responsive site] with [n] primary sections and [n] secondary items.
Task: Recommend and specify a mobile navigation pattern (bottom tabs / hamburger / tab bar + more).
Context: Core user tasks: [list]. Frequency of each task: [high/medium/low]. OS conventions: [iOS HIG / Material]. One-handed reachability: [required/preferred].
Output: Pattern recommendation with rationale. Specification including: icon + label for each item, overflow handling, active/inactive states, and gesture support. Include ARIA landmarks.

6. Dashboard Layout

Situation: Designing a data dashboard for [user role] displaying [n] widgets/data types.
Task: Create a dashboard layout specification with widget hierarchy and responsive behaviour.
Context: Primary KPIs: [list]. Update frequency: [real-time/periodic]. Customisation level: [fixed/rearrangeable/fully custom]. Accessibility: data tables as alternatives to charts.
Output: Dashboard grid layout per breakpoint. Widget list with: priority, default size, minimum size, and data visualisation type. Include keyboard navigation pattern.

7. Form Flow Wireframe

Situation: Designing a [multi-step/single-page] form for [purpose] with [n] fields.
Task: Specify form layout, field grouping, validation strategy, and error handling.
Context: Required fields: [list]. Optional fields: [list]. Conditional logic: [description]. Target completion rate: [percentage]. Accessibility: visible labels, error association with aria-describedby.
Output: Form specification with: field groups, layout per breakpoint, validation rules (inline/on-submit), error message patterns, and progress indication for multi-step forms.

8. Accessibility-First Layout

Situation: Designing [page/component] with accessibility as the primary design constraint.
Task: Create a layout specification that meets WCAG 2.2 AA and addresses [specific disabilities, e.g. low vision, motor impairment].
Context: Assistive technologies to support: [screen readers, switch access, voice control, magnification]. Content reflow requirement: up to 400% zoom. Colour palette: [colours with contrast ratios].
Output: Layout specification with: semantic HTML structure, heading hierarchy, landmark regions, focus order, skip links, touch target sizes, and alternative interaction methods.

UX Writing & Microcopy: 8 Templates

Prompts that specify tone-of-voice guidelines and character limits produce microcopy scoring 39% higher on clarity metrics (AI Prompt Architect platform data, Q2 2026). Including the user's emotional state in error message prompts produces outputs rated 2.6x more empathetic by reviewers. These patterns are consistent with what we observe across business prompting use cases more broadly.

WCAG Note: All microcopy templates include prompts for plain language (reading level), meaningful alt-text, and screen-reader-friendly phrasing. Error messages must be programmatically associated with the relevant input field.

1. Error Message System

Situation: Designing a systematic error message framework for [product].
Task: Write error messages for [n] error types covering [validation, system, permission, connectivity] categories.
Context: Tone: [supportive/neutral/technical]. User emotional state at error point: [frustrated/confused/anxious]. Character limit: [n]. Brand voice: [description]. Accessibility: messages must be announced by screen readers via aria-live regions.
Output: Error message table with: Error Code, Category, User-Facing Message, Recovery Action, and Tone Notes. Include one positive and one negative example per category.

2. Onboarding Copy

Situation: Writing onboarding copy for [product] targeting [user segment] with [experience level].
Task: Create copy for a [n]-step onboarding flow covering [setup, feature discovery, first value moment].
Context: Key activation metric: [description]. Average time to first value: [current duration]. Competitor onboarding benchmark: [description]. Reading level: [grade].
Output: Copy for each step with: headline (max [n] characters), body text (max [n] words), CTA label, and tooltip/helper text. Include skip option copy.

3. CTA Optimisation

Situation: Optimising CTAs for [page/flow] with current conversion rate of [percentage].
Task: Write [n] CTA variants for A/B testing across [primary, secondary, tertiary] button tiers.
Context: User intent at this point: [browsing/comparing/ready to act]. Value proposition: [description]. Character limit: [n]. Accessibility: CTAs must describe their action (avoid "Click here").
Output: CTA variants table with: Variant, Button Text, Supporting Text, Hypothesis, and Expected Lift. Include rationale for word choice.

4. Empty State Copy

Situation: Writing empty state messages for [n] screens/components in [product].
Task: Create empty state copy that guides users toward their first action.
Context: Component types: [list, dashboard, inbox, search results, etc.]. User stage: [new/returning]. Brand tone: [description]. Illustration style: [if applicable].
Output: Copy per component with: headline, body text, CTA, and illustration brief. Distinguish between "no data yet" and "no results found" states.

5. Tooltip Content

Situation: Writing tooltips for [n] UI elements in [product/feature].
Task: Create tooltip copy that explains functionality without interrupting flow.
Context: Tooltip trigger: [hover/focus/click]. Max character count: [n]. User expertise: [beginner/intermediate/expert]. Accessibility: tooltips must be keyboard-accessible and dismissible.
Output: Tooltip table with: Element, Trigger, Tooltip Text, and Progressive Disclosure Notes (what detail to show on expanded help).

6. Confirmation Dialogue

Situation: Writing confirmation dialogues for [destructive/irreversible/high-stakes] actions in [product].
Task: Create confirmation dialogue copy that prevents accidental actions while reducing friction for intentional ones.
Context: Actions requiring confirmation: [list]. Undo capability: [yes/no per action]. User emotional state: [confident/uncertain]. Accessibility: dialogue must trap focus and be dismissible via Escape key.
Output: Dialogue copy per action with: Title, Body Text, Confirm Button, Cancel Button, and Additional Context. Include one variant for power users (reduced friction).

7. Loading State Messages

Situation: Writing loading, progress, and wait-state messages for [product] where [operations] take [duration range].
Task: Create loading state copy that manages user expectations and reduces perceived wait time.
Context: Loading types: [skeleton/spinner/progress bar/percentage]. Average wait: [duration]. Maximum wait: [duration]. Brand tone: [description].
Output: Loading messages per type with: Initial Message, Extended Wait Message (after [n] seconds), and Failure/Timeout Message. Include animation direction notes.

8. Accessibility Alt-Text

Situation: Writing alt-text for [n] images/icons/illustrations in [product/page].
Task: Create descriptive, functional alt-text following WCAG 2.2 guidelines.
Context: Image types: [decorative/informative/functional/complex]. Screen-reader usage in user base: [percentage]. Include long descriptions for complex images (charts, infographics). Decorative images should have alt="".
Output: Alt-text table with: Image ID, Image Type, Alt-Text, Long Description (if needed), and Reasoning. Flag images that should be marked decorative.

Usability Testing & Heuristic Evaluation: 8 Templates

Referencing Nielsen's heuristics by number in evaluation prompts produces findings 47% more granular than generic "evaluate usability" requests (AI Prompt Architect platform data, Q2 2026). Prompts specifying task-completion-rate targets generate success criteria 2.3x more measurable. Both findings hold across moderated and unmoderated study types.

1. Moderated Test Script

Situation: Conducting moderated usability testing with [n] participants for [product/feature].
Task: Write a moderated test script with [n] tasks, think-aloud protocol, and post-task questionnaires.
Context: Tasks map to user stories: [list]. Success criteria: [task completion rate target]. Session length: [duration]. Recording: [screen + audio/video]. Participant incentive: [type].
Output: Script with: Introduction (consent, think-aloud instructions), Tasks (numbered with scenario, success criteria, time limit), Post-task questions (SEQ or similar), and Debrief questions.

2. Unmoderated Test Plan

Situation: Running unmoderated remote usability testing via [platform, e.g. UserTesting, Maze] with [n] participants.
Task: Create an unmoderated test plan including task design, screener questions, and success metrics.
Context: Target demographics: [description]. Screener criteria: [list]. Tasks must be self-explanatory. Maximum test duration: [minutes].
Output: Test plan with: Screener questions, Task list (with written scenarios), Success metrics per task, and Post-test survey. Include pilot test checklist.

3. Heuristic Evaluation Checklist

Situation: Evaluating [product/feature] against Nielsen's 10 usability heuristics.
Task: Generate a heuristic evaluation checklist with specific sub-criteria for [product type, e.g. e-commerce, SaaS dashboard, mobile app].
Context: Evaluators: [n, expertise level]. Scope: [specific flows/pages]. Severity rating scale: [0–4 Nielsen scale]. Include WCAG 2.2 AA criteria mapped to relevant heuristics.
Output: Checklist table with: Heuristic Number, Heuristic Name, Sub-Criterion, Location, Severity, and Recommendation. Include a severity distribution summary.

4. Cognitive Walkthrough

Situation: Performing a cognitive walkthrough of [task flow] for [user type] in [product].
Task: Walk through each step answering: Will the user try the right action? Will they notice the correct control? Will they understand the feedback?
Context: User's goal: [description]. Prior knowledge assumed: [list]. Number of steps in flow: [n]. Include accessibility barriers at each step.
Output: Step-by-step walkthrough table with: Step Number, Action Required, Success Story, Failure Story, and Design Recommendation.

5. A/B Test Hypothesis

Situation: Designing A/B tests for [n] UX changes in [product].
Task: Formulate structured hypotheses in the format: "If we [change], then [metric] will [direction] by [amount] because [rationale]."
Context: Current metrics: [baselines]. Changes under consideration: [list]. Traffic available: [monthly users]. Test duration constraints: [minimum/maximum days]. Statistical significance target: [e.g. 95%].
Output: Hypothesis table with: Change, Primary Metric, Expected Lift, Required Sample Size, Test Duration, and Risk Assessment.

6. Accessibility Audit (WCAG 2.2)

Situation: Conducting a comprehensive WCAG 2.2 [A/AA/AAA] audit for [product].
Task: Evaluate against all applicable success criteria and generate a prioritised remediation plan.
Context: Audit scope: [pages/flows]. Assistive technologies: [NVDA, JAWS, VoiceOver, TalkBack, Switch Access]. Automated scan results from [tool, e.g. axe, WAVE]: [summary]. Known issues: [list].
Output: Audit report with: Criterion, Level, Status, Impact (users affected), Effort (dev hours estimate), and Fix Description. Prioritise by impact × effort matrix.

7. SUS Analysis

Situation: Analysing System Usability Scale (SUS) results from [n] respondents for [product].
Task: Calculate SUS scores and provide interpretive analysis with benchmarking.
Context: Raw SUS responses: [data or summary statistics]. Industry benchmark: [score]. Previous SUS score: [score, if available]. Test conditions: [description].
Output: Analysis with: Overall SUS score, Grade (A–F), Percentile rank, Learnability sub-score, Usability sub-score, comparison to benchmarks, and Recommended focus areas for improvement.

8. Competitive Usability Benchmark

Situation: Benchmarking [product] usability against [n] competitors for [specific tasks].
Task: Design a competitive usability benchmark study measuring [task completion, time-on-task, errors, satisfaction].
Context: Competitors: [list]. Tasks to benchmark: [list, matching real user scenarios]. Participant criteria: [description]. Budget: [participant count × sessions].
Output: Benchmark study plan with: Task matrix, Metrics per task, Statistical analysis approach, Reporting template, and Competitor comparison dashboard specification.

Prompt Chaining for End-to-End UX Workflows

Individual prompts produce individual deliverables. Real UX work requires connected deliverables — research that feeds personas, personas that inform IA, IA that shapes wireframes, wireframes that guide testing. Prompt chaining closes this loop.

Here is a five-stage chain using templates from this guide:

The key technique: include the previous stage's output verbatim in the Context field of the next prompt. This preserves specificity and ensures each deliverable builds on validated data rather than generic assumptions. Our platform data shows that chained prompts retain 89% of the specificity from the initial research stage through to the testing stage, compared to 34% when each prompt is written independently (AI Prompt Architect platform data, Q2 2026).

For design patterns around prompt chaining and template composition, see our guide on prompt engineering best practices.

How to Score & Optimise Your UX Prompts

After one STCO-guided iteration, prompts improve by a median of 19 points on a 100-point scale (n=3,100, AI Prompt Architect platform data, Q2 2026). The Prompt Scorer tool on our platform evaluates prompts across four dimensions — Specificity, Structure, Context Completeness, and Output Clarity — and provides line-by-line improvement suggestions.

A practical optimisation workflow:

• Step 1: Draft your prompt using the relevant template from this guide.
• Step 2: Run it through the Prompt Scorer. Note scores below 70 on any dimension.
• Step 3: Add missing STCO elements — most commonly, the Context field lacks user demographics or technical constraints.
• Step 4: Re-score. Target 80+ across all four dimensions before using the prompt in production.
• Step 5: Save high-scoring prompts to your template library for team reuse.

The most common gap in UX prompts is missing accessibility context. Adding WCAG level, assistive technology requirements, and colour contrast ratios typically lifts the Specificity score by 12–18 points alone.

Further Reading

• The STCO Framework Guide
• AI Prompting Tools: Complete Guide
• Prompt Engineering Best Practices
• AI Prompts for Business: Complete Guide
• Prompt Template Design Patterns
• AI Prompt Best Practices