Review evidence, outcomes, and team practices without turning the meeting into a blame session. Use the example as a starting point, then adapt the steps to your product, team, and evidence.
WHAT YOU’LL LEAVE WITH
A concise retrospective with two or three owned improvements.
Before you begin
- The original goal and decision record
- Delivery and outcome evidence
- Input from the people who did and supported the work
STEP 01
Revisit the original intent
Start with the problem, expected outcome, and assumptions recorded before the work. Separate what was known at the time from information learned later. Avoid judging a past decision only by its eventual outcome.
STEP 02
Build a factual timeline
Record important changes, discoveries, handoffs, and constraints. Use source material and invite corrections. AI can organize approved notes, but people should verify the sequence and interpretation.
STEP 03
Examine delivery and outcomes
Discuss what shipped and what happened for users. A delivery delay and a failed customer hypothesis need different responses. Look for measurement gaps before drawing confident conclusions.
STEP 04
Understand the mechanisms
Ask what conditions contributed to the result. Examine incentives, missing context, capacity, tooling, and decision processes. Do not reduce a complex failure to a single person or an unsupported “root cause.”
STEP 05
Choose a small set of changes
Select two or three improvements with clear owners and observable completion. “Communicate better” is vague. “Add a named support reviewer before launch approval” is something the team can try.
STEP 06
Check whether it helped
Schedule a follow-up and review the changes during the next relevant project. Keep useful practices and revise ineffective ones. A retrospective is complete when learning changes behavior, not when the notes are filed.
MAKE THIS PROMPT YOUR OWN
Organize these approved retrospective notes into facts, interpretations, unresolved questions, and proposed improvements. Preserve different perspectives. Do not assign blame or infer motives. Suggest a small number of observable actions, leaving owner assignments for the team.
Check your work
- Are facts separate from hindsight and interpretation?
- Are product outcomes discussed alongside delivery?
- Are actions specific and owned?
- Is there a follow-up to check their effect?
Further reading: related templates & resources ↗
Editorial guide · Reviewed September 27, 2026. Tool interfaces may change.