How to run an agile retrospective: the complete guide

A retrospective is when the team looks at how it works, not just at what it delivered. Run well, it is the most profitable meeting of the sprint: one hour to fix what costs dozens. Run badly, it quickly becomes the one everyone would rather skip. Three symptoms almost always give it away: the session overruns, three people talk while five stay silent, and the action items decided last time were never read again.

This guide covers the flow that has proven itself for twenty years, the role of the person who facilitates, and the traps that make a retro go through the motions.

What is an agile retrospective for?

The last of the twelve principles of the Agile Manifesto sums it up in one sentence: “At regular intervals, the team reflects on how to become more effective, then tunes and adjusts its behavior accordingly.” It is all in the end of the sentence. A retro that produces a good discussion but no change has not done its job.

In Scrum, the sprint retrospective closes every sprint, after the review. The Scrum Guide caps it at three hours for a one-month sprint, and less for shorter sprints. But nothing requires Scrum to hold one: a Kanban team, a support team or a leadership team gains just as much from taking this step back at a fixed rhythm, every two or three weeks.

The five steps of a retrospective

In Agile Retrospectives, Esther Derby and Diana Larsen formalized a five-step flow that has become the reference. For a one-hour session, here is how to split it.

0 min60 min

1. 1. Open 3 mingoal, ground rules, one word each
2. 2. Gather 10 minwrite before you talk
3. 3. Understand 30 mingroup, vote, dig in
4. 4. Decide 12 minone action item per topic, with an owner and a date
5. 5. Close 5 minread back the action items, feedback on the session

1. Open the session (2 to 3 minutes)

Restate the goal, the time available and the ground rules, then have everyone speak once, even a single word. This step is short, but it is not skipped: someone who has said nothing in the first few minutes has a good chance of saying nothing at all. A mood check or a short question (“the sprint in one word?”) is enough.

2. Gather the facts (8 to 10 minutes)

Everyone writes down what they experienced during the period, one idea per card, in the columns of the chosen format. The decisive point is to write before you talk, preferably without seeing other people’s cards: otherwise the first cards on the board steer all the ones after them, and the retro merely confirms the opinion of the fastest writers. No need to drag it out: after ten minutes or so, new cards become rare, and every minute saved here goes to the discussion.

3. Understand (25 to 30 minutes)

Group the cards that talk about the same thing, then the team votes to pick the two or three topics worth digging into. The vote is what protects the session from overrunning: you don’t discuss everything, you discuss what matters most to the group. On each topic, look for the cause rather than the culprit. This is the heart of the session, and the step that gets the most time: a retro that gathers a lot but discusses little produces lists, not decisions.

1. 1Specs arrive the day before the sprint7
2. 2Too many interruptions at the end of the sprint5

3. if time allows

   3A meeting-free half-day3
4. 4The dashboard is too slow1

4. Decide (10 to 15 minutes)

Each topic discussed yields at most one action item, with a named owner and a due date. One or two action items that get done beat six forgotten ones. This is the step most often rushed, because it comes at the end of the session, when time is short: protect it. A whole article is devoted to it: why retro action items never get done.

5. Close (5 minutes)

Read the action items out loud, say thank you, and ask for feedback on the session itself: what would have made it more useful? A retrospective improves through retrospectives too.

How much time to plan?

A common rule of thumb: about 45 minutes per week of sprint, so an hour and a half for a two-week sprint. In practice, many teams run a very good retro in one hour, provided each step is timed. What matters is not the total length but that it is announced and kept: a session that always overruns teaches the team that the end, and therefore the action items, are optional.

The facilitator’s role

The facilitator holds the frame, not the content. They move the steps along, keep time, share out the floor and bring people back to the topic; they don’t settle debates or defend their own cards. That is why it is better if they are not always the team’s manager, whose presence weighs on what gets said.

Rotating facilitation among team members has two benefits: everyone learns to run a session, and the format renews itself, since each facilitator has their own habits.

The Prime Directive

Many teams open their retro by reading Norm Kerth’s “Prime Directive”:

“Regardless of what we discover, we understand and truly believe that everyone did the best job they could, given what they knew at the time, their skills and abilities, the resources available, and the situation at hand.”

It is a reminder that you are examining a system, not people. A retro that turns into settling scores stops producing anything: people stop saying what is wrong.

The most common traps

• Always the same format. After ten identical sessions, the cards write themselves. Varying the columns makes the team look at the sprint from another angle: see six retrospective formats and when to use them.
• Discussion before the vote. Commenting on each card as it is posted guarantees you never finish.
• Too many action items. A list of eight action items is a wish list.
• Action items nobody revisits. If the next retro doesn’t start with the status of the previous one’s action items, the team quickly learns they commit no one to anything.
• Topics out of reach. “Management should…” is not an action item. Look for what the team can change itself, or who it can go and talk to.

If your retro shows several of these symptoms at once, read how to revive a retrospective that is running out of steam.

And with SquadRetro?

SquadRetro follows this flow: brainstorm, grouping, vote, discussion, closing. The facilitator moves the phases along and the whole team’s screens follow at the same moment. During the brainstorm, everyone sees only their own cards, which stops people copying their neighbor. In discussion, cards come up in vote order, and the facilitator presents them one by one. Every action item leaves the session with an owner and a due date, and stays available on the squad’s page between retros.