Retrospective Meeting Agenda

What success looks like

Turn recent experience into a small number of owned improvements. The practical test is whether the people involved can point to specific learning and one or two experiments without reconstructing the whole conversation.

A practical method

Set the boundary for the work

Write specific learning and one or two experiments as the boundary. Keep only the evidence, authority, and participation needed to produce that specific result.

Work through the sequence

Set the scope, gather observations silently, group patterns, explore causes without blame, choose experiments, and assign review dates. Place that instruction in the shared workspace and use it to distinguish the real work from background discussion.

Create a usable record

A durable record should make sense beside this example: The team tests a two-day review window for one sprint and checks whether blocked work falls. Name any owner, due point, decision rule, or review trigger needed for another person to act.

Worked example

The team tests a two-day review window for one sprint and checks whether blocked work falls.

Copy this example

The example connects directly to specific learning and one or two experiments. Put similarly concrete wording in the invitation, agenda, or record so the result can be checked later.

For a related example aimed at a lighter recurring meeting system, continue with How to Audit Recurring Meetings.

A failure to avoid

Watch for Generating a long improvement list with no owner, limit, or review point. If it appears, pause and compare the discussion with specific learning and one or two experiments before more meeting time is used.

A useful recovery question

Ask: “What is the smallest responsible move that would still create specific learning and one or two experiments?” If the answer depends on missing evidence, record that evidence, one owner, and the exact return condition.

Final checklist

• The stated goal is specific learning and one or two experiments.
• The working method is visible to everyone involved.
• The record is specific enough to support an example such as: The team tests a two-day review window for one sprint and checks whether blocked work falls.
• The design actively guards against Generating a long improvement list with no owner, limit, or review point.

Next, use the meeting agenda builder to turn the method into a working artifact.