User Research

User Research

Overview

User research bridges the gap between what we think users need and what they actually need. This skill covers the full research lifecycle — from planning and conducting studies to synthesizing findings and making them actionable. Works for both qualitative (interviews, diary studies, field visits) and quantitative (surveys, analytics, logs) approaches.

When to Use

• Starting a new initiative and need to understand user needs
• Making a product decision where the cost of being wrong is high
• Building a research repository for your team
• Validating or invalidating assumptions about user behavior
• Don't use for: trivial UI decisions (A/B test or gut check is faster), already-well-understood problems

Instructions

Phase 1: Plan the Research

1. Define the research question — What decision will this research inform? Frame as: "We need to decide X. What do we need to learn to make that decision confidently?"
2. Choose the method:

• Generative (uncover unknown needs): interviews, diary studies, field observation
• Evaluative (test a concept): usability tests, concept tests, A/B tests
• Descriptive (measure behavior): surveys, analytics, log analysis

1. Recruit participants — Define screener criteria. Aim for 5-8 per segment for qualitative, 200+ for surveys

Phase 2: Conduct the Research

For qualitative:

1. Follow the customer-interviews skill for unstructured interviews
2. For usability tests: give tasks, don't guide. Measure success rate, time-on-task, and satisfaction
3. Take detailed notes. Record (with permission). Use a second observer if possible

For quantitative:

1. Design surveys to avoid bias (no leading questions, balanced scales, randomize order)
2. Use analytics to measure actual behavior, not self-reported behavior
3. Triangulate: survey findings should be validated against behavioral data

Phase 3: Synthesize Findings

1. Affinity mapping — Group observations by theme on a virtual or physical wall
2. Thematic analysis — Identify 5-7 core themes. Each theme should have 3+ supporting data points
3. Create research artifacts:

• Insight statements: "We learned X, which means Y, so we should do Z"
• Journey maps showing pain points
• Opportunity areas ranked by evidence strength

1. Rate confidence — How many sources support each finding? High (5+ sources), Medium (3-4), Low (1-2)

Phase 4: Make it Actionable

1. Write recommendations — Connect each finding to a specific action or decision
2. Present findings — Use the "Headline, Evidence, Implication, Action" format
3. Store in research repo — Tag by topic, segment, and date. Make it searchable
4. Track impact — Did the recommendation lead to a decision? Measure research ROI

Sample Output

A research synthesis for a fintech budgeting feature might look like:

Theme: Users overestimate their self-control

• Evidence: 6/8 participants said they "budget fine" but analytics showed 80% overshoot by 20%+
• Implication: Willpower-based budgeting fails. Users need hard constraints, not tracking
• Recommendation: Add spending limits (not just tracking). Auto-pause when limit is hit
• Confidence: High (8 interviews + 3 months analytics)

Common Pitfalls

1. Research without a decision — If your research doesn't inform a decision, don't do it
2. Cherry-picking — Only reporting findings that support your hypothesis
3. Over-generalizing — 5 interviews don't speak for all users. Segment your findings
4. Analysis paralysis — More research isn't always better. Find the 80% answer and act
5. Not sharing findings — Research that lives in a doc no one reads never provides value

Verification Checklist

• [ ] Research question tied to a specific decision
• [ ] Method chosen matches the question type (generative/evaluative/descriptive)
• [ ] At least 5 participants per segment for qualitative
• [ ] Affinity mapping completed with 5+ themes
• [ ] Each finding rated for confidence level
• [ ] Recommendations given for each key finding
• [ ] Findings stored in a searchable format

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: uthumany
• Source: uthumany/uthy-legacy-os
• License: MIT
• Homepage: https://uthumany.github.io/uthy-legacy-os/

Install and usage instructions live in the source repository linked above.