Turn research into priorities by preserving source observations, writing findings that explain behavior and consequence, and comparing them against the product decision. Consider severity, reach evidence, strategic relevance, confidence, dependencies, and cost of delay without hiding uncertainty in a score. Record what is selected, deferred, or rejected and why.
Who this is for: Designers and product teams converting interviews, usability sessions, support evidence, and analytics into a focused design plan.
- Keep a traceable chain from raw observation to finding, implication, opportunity, and decision.
- Compare consequence, evidence strength, affected context, strategy, and dependencies rather than quotation volume.
- Publish selected and deferred priorities with owners, rationale, limits, and the next evidence check.
Normalize the evidence
Gather notes, recordings, support cases, analytics, and existing decisions under their consent and retention rules. Label participant or source, context, event, date, and exact location. Separate what happened from the team's interpretation so later reviewers can challenge reasoning without erasing the underlying observation.
Create evidence units with situation, goal, action, obstacle, consequence, and relevant quotation. Retain contradictions and successful cases. A participant's vivid complaint can expose an important mechanism, but it does not by itself establish how often the problem occurs across the customer population.
Write findings with consequence
A useful finding explains a pattern and why it matters. 'Reviewers reopened the attachment because the approval screen did not show its revision date, creating uncertainty about which file they approved' is more actionable than 'metadata is confusing.' Link every claim to the evidence units that support and complicate it.
Distinguish a finding from a solution request. Participants may ask for color coding, but the evidence may show difficulty distinguishing status in several contexts. Preserve the underlying need so designers can compare text, grouping, order, interaction, and process responses rather than inheriting one requested treatment.
Compare priorities explicitly
Discuss customer consequence, tasks affected, evidence strength, likely reach, strategic outcome, risk, reversibility, and dependencies. Use ranges or qualitative judgments when data is thin. A severe access failure affecting a smaller known group may deserve action before a common cosmetic irritation, depending on obligations and consequence.
Do not average away legal, privacy, security, or accessibility requirements inside a weighted score. Confirm such constraints with accountable specialists. For discretionary findings, a matrix can structure discussion, but the decision record should explain why the chosen criteria matter and where confidence remains limited.
Move from priority to action
For each selected area, state the outcome, design question, owner, next method, and decision date. Some findings justify direct repair; others require concept exploration, workflow research, technical investigation, or measurement. Keep the proposed response proportional to the evidence instead of turning every finding into a roadmap feature.
Share deferred findings and reasons with stakeholders, then define review triggers such as new evidence, a related launch, or a changed obligation. Later, verify whether the design response altered the observed behavior without creating new harm. Closing a design task is not evidence that the original problem was resolved.
Prioritize hypothetical onboarding findings
Hypothetical research finds unclear account roles, a hard-to-notice optional avatar step, and occasional loss of entered tax data.
- Trace each finding to observed sessions, support records, affected context, and contradictory evidence.
- Compare consequence and obligations, recognizing that lost tax data has greater recovery cost than the avatar issue.
- Select data preservation for immediate repair, investigate role language next, and defer avatar prominence with rationale.
- Assign owners and verify the revised tax flow with failure-state tests before calling the issue resolved.
Finding priority record
Use one record for each decision-relevant research finding.
- Finding, context, affected goal, mechanism, customer consequence, and contradictory cases.
- Evidence links, source mix, observed sample, behavioral data, confidence, and prevalence limits.
- Decision factors: severity, reach evidence, strategy, obligation, dependency, reversibility, and delay cost.
- Response options, selected action, deferred alternatives, owner, rationale, and stakeholder communication.
- Verification method, guardrails, review date, trigger, observed result, and remaining uncertainty.
Common mistakes
- Counting sticky notes or mentions and treating the largest cluster as automatically most important.
- Converting requested features directly into priorities without preserving the underlying behavior and need.
- Publishing only selected work and losing the evidence and rationale behind deferred findings.
Try one
Five participants dislike an icon, while one keyboard participant cannot reach the related action. Which finding should lead?
A strong answer rejects priority by mention count alone. The keyboard barrier blocks completion and invokes a specific access concern, so it requires prompt criterion-level evaluation and likely repair. The icon reaction still needs context and consequence before prioritization. The response should cite evidence limits, accountable accessibility review, and verification rather than claiming one session represents every user.
Sources
- Nielsen Norman Group qualitative analysis guideGuidance on organizing observations and building defensible themes from qualitative research.
- Nielsen Norman Group affinity diagram guideGuidance on grouping research observations while preserving their source and meaning.