USEFUL AI BEATS HYPE
Create a product-management skill that turns customer feedback into an evidence brief
An evidence brief with themes, source links, hypotheses and one proposed next experiment.
By Ashley Kays. Worked examples use fictional inputs; expected results are illustrations for testing, not performance claims.
A request is evidence of a problem worth investigating. It is not automatically a roadmap commitment. This skill preserves the customer signal while helping a product manager compare possible responses and identify the next question to answer.
What you will build
An evidence brief with themes, source links, hypotheses and one proposed next experiment.
Gather the inputs
Feedback excerpts with anonymous IDs; customer segment if known; dates; current product goals; known constraints; existing research. Remove unnecessary personal information.
1. Define the decision
Choose a question such as which onboarding problem to investigate next. Avoid asking the skill to create the entire roadmap.
2. Separate observation from interpretation
Have it preserve what people said and label its explanations as hypotheses.
3. Group without inflating evidence
Count distinct supplied sources. Keep duplicate excerpts from the same source from becoming extra votes.
4. Connect to the product goal
Explain why each theme may matter to the stated outcome. Leave impact and effort unknown when evidence is missing.
5. Choose a learning step
Propose a small experiment, its success measure and the assumption it tests. Keep the prioritization decision with the product owner.
A worked example
Fictional input: C1 cannot tell whether setup was saved. C2 reports the same issue. C3 is a second quote from C1. Product goal: improve first-run completion. No analytics or engineering estimate supplied.
Theme: uncertainty about setup status. Evidence: two distinct sources, C1 and C2; C3 is not an additional customer. Hypothesis: a clearer saved-state message may help. Next experiment: test whether people can correctly identify completion in a prototype. Impact size and engineering effort: unknown.
Copyable skill instructions
Use this focused template in the skill-creation flow described in the starter guide. Review and adapt it to your own work before saving. It is not installed by downloading this article.
---
name: feedback-evidence-brief
description: >-
Organize supplied customer feedback into an evidence brief for product discovery. Use
before prioritization, not for making roadmap commitments.
---
# Feedback to evidence brief
Require feedback and the decision or product goal. Preserve source IDs and dates. Separate observations, interpretations and hypotheses. Deduplicate by source when identity is known; flag uncertainty otherwise. Return themes, source count, supporting excerpts, counterevidence, unknowns and one proposed experiment. Do not fabricate research, market size, effort or impact scores. Do not create roadmap promises or edit the backlog without a separate instruction.
Required output: An evidence brief with themes, source links, hypotheses and one proposed next experiment.Three tests to run
| Test prompt | Pass condition |
|---|---|
| Normal: five labeled customer notes | Themes remain linked to the exact sources. |
| Missing: no reach, impact or effort estimates | Unknowns remain unknown; no invented prioritization score. |
| Conflict: customers want opposite behaviors | Both needs appear, with segment or context questions. |
How to judge whether it helps
Track time to an evidence brief, unsupported claims removed and source traceability. Check whether the brief improves the next discovery decision, not merely whether it looks organized.
What to build next
Next skills: interview synthesis, PRD readiness review, acceptance-criteria drafting, release-note drafting.
Return to the illustrated setup guide and official sources.
Bring a real task.
Leave with a working AI-assisted process. Ashley and the Waymaker team can help you choose the workflow, create the skill, test it and make it useful for your team.
Book a 15-minute fit call