Retrospective template: 4 formats and an agenda that works
Free sprint retrospective templates: went well and didn't, 4 Ls, sad mad glad, and start stop continue, with a timed agenda and action items that stick.
The team holds a retrospective every two weeks. The same three complaints come up each time: flaky tests, late requirements, too many meetings. Someone writes them on sticky notes, everyone agrees, and nothing changes, because nobody owned a single action.
A retrospective is only as useful as the change that follows it. This guide gives four formats and an agenda, from the Scrum Guide and Atlassian's retrospective play, checked on 16 November 2026.
What it's for
The Scrum Guide describes the sprint retrospective as planning ways to increase quality and effectiveness. The team inspects how the sprint went across individuals, interactions, processes, tools, and its definition of done; identifies assumptions that led it astray; and discusses what went well, what problems came up, and how they were or weren't solved. It then picks the most helpful changes, and the most impactful are addressed as soon as possible, sometimes in the next sprint's backlog.
It's timeboxed to three hours for a one-month sprint, and usually shorter for shorter sprints.
The agenda
For a one-hour retro on a two-week sprint:
| Step | Time | What happens |
|---|---|---|
| 1. Set the tone | 5 min | Restate that it's blameless and where notes go |
| 2. Review last actions | 5 min | Did the changes we chose last time happen? |
| 3. Gather feedback | 15 min | Use one of the formats below |
| 4. Find themes | 15 min | Group similar points, vote on what matters most |
| 5. Decide actions | 15 min | One to three changes, each with an owner and date |
| 6. Close | 5 min | Read back the actions |
Atlassian's play stresses step 1: encourage honest input focused on improvement, not blame, and say whether the discussion stays in the room.
Four formats
Went well, didn't go well, change
| Went well | Didn't go well | What we'll change |
|---|---|---|
The default. Atlassian lists the core questions as what went well, what went wrong, what you learned, and what to change.
4 Ls
| Loved | Loathed | Learned | Longed for |
|---|---|---|---|
Good for drawing out learning and wishes, not only complaints.
Sad, mad, glad
| Sad | Mad | Glad |
|---|---|---|
Atlassian suggests it when feedback is thin; it gives people a way into how the sprint felt.
Start, stop, continue
| Start | Stop | Continue |
|---|---|---|
Goes straight to behavior. Good when the team already knows the problems and needs to choose.
The action items template
The part that decides whether anything changes:
## Retro actions: Sprint 24
| Action | Owner | Due | Status |
|---|---|---|---|
| Quarantine the 6 flakiest tests and file tickets | Arjun | Sprint 25, day 3 | |
| PM shares acceptance criteria 2 days before planning | Priya | Next planning | |
| Cancel Thursday sync; move updates to the channel | Mei | This week | |
## Carried over from last retro
| Action | Owner | Done? |
|---|---|---|
Rules that help:
- One to three actions. Ten actions means none.
- An owner each, not "the team."
- In the backlog, so they compete for time like other work.
- Reviewed first at the next retro.
Common problems
- Same complaints every time. No owned actions, or actions too big to finish. Make them smaller.
- Quiet room. Try anonymous input or sad, mad, glad. Atlassian suggests anonymous submissions when feedback is scarce.
- Focusing on one-off events. Atlassian advises looking at what affects the next sprint, not isolated mistakes.
- Blame. Shut it down early, or people stop raising issues.
Retro outcomes are decisions
"PM shares criteria two days before planning" and "we stop the Thursday sync" are team decisions. Written only on a retro board, they fade by the next sprint. Recorded with an owner, they stick, and the people they affect, including anyone new to the team, can find them.
Where Lockstep fits
Lockstep captures decisions like these from documents, Slack, and pull requests, routes each to its owner for approval, and keeps them where engineers and Claude Code sessions are briefed on the ones that apply to their work.
See Lockstep for product teams.
Sources checked 16 November 2026
- https://scrumguides.org/scrum-guide.html
- https://www.atlassian.com/team-playbook/plays/retrospective
Related reading
For incident reviews, see the blameless postmortem template. For planning the next sprint, see sprint planning with AI. For other team meetings, see the meeting notes template.
Frequently asked questions
- What is a sprint retrospective?
- The Scrum Guide describes it as the event where the team plans ways to increase quality and effectiveness. It inspects how the last sprint went across people, interactions, processes, tools, and the definition of done, and picks the most helpful changes, which may go into the next sprint backlog.
- How long should a retrospective be?
- The Scrum Guide sets a maximum of three hours for a one-month sprint, usually shorter for shorter sprints. Atlassian's guide says retrospectives can be as short as 45 minutes. For a two-week sprint, an hour is common.
- What questions should a retrospective ask?
- Atlassian suggests: What went well? What went wrong? What did you learn? What changes can we make? Formats like the 4 Ls (loved, loathed, learned, longed for) or sad, mad, glad give the same questions a structure.
- How do I make retrospective action items actually happen?
- Pick one to three, give each an owner and a date, put them in the next sprint backlog, and start the next retrospective by reviewing them. The Scrum Guide says the most impactful improvements should be addressed as soon as possible.
- Should retrospectives be blameless?
- Yes. Atlassian's play advises focusing on improvement instead of blame and making clear where the discussion will be shared. People hold back problems if they expect blame.
Keep reading