Starfish Retrospective

What is the Starfish retrospective?

The Starfish retrospective sorts the team's practices into five categories instead of the usual three: keep doing, more of, less of, start doing and stop doing. It was introduced by Patrick Kua as a refinement of Start Stop Continue, and the refinement is the whole point — two extra categories for degree, so a team can say "this is useful but there is too much of it" without having to argue that it should be abolished.

That distinction matters more than it sounds. Most disagreements in a retrospective are not about whether a practice is good; they are about how much of it there should be. Given only start, stop and continue, those disagreements collapse into a fight, and the usual outcome is that nothing changes. Starfish gives them a column.

When to use the Starfish retrospective

• When the team knows its problems and needs to agree on what to change rather than discover what is wrong.
• When Start Stop Continue keeps producing the same three cards sprint after sprint with no movement.
• When a practice is contested — a heavy process, long standups, thorough documentation — and the honest position is "less", not "stop".
• A few sprints after adopting a new process or tool, to tune it rather than reopen the decision.
• When the team is doing well and the useful conversation is about what to protect and what to amplify.
• For a quarterly review of ways of working, where five categories give enough resolution to be worth the hour.

It is the wrong format when the team does not yet understand what happened. Starfish starts from proposals, so a confusing or painful sprint is better served by a discovery format first — Mad Sad Glad, 4Ls or a timeline — with Starfish run afterwards to decide.

The columns explained

Keep doing

Practices that are working at roughly the right level. This is a protective column: the habits a team drops first under deadline pressure are the ones nobody ever wrote down as valuable. Naming them makes them harder to lose quietly.

More of

Things the team already does occasionally and should do routinely. Pairing that only happens during incidents, documentation written only on request, testing that stops at the happy path. Nothing new has to be invented here — the practice exists and simply needs a bigger share of the week.

Less of

Things that are useful in smaller doses. Meetings that could be half the length, work in progress spread across too many branches, alerts that fire without anyone acting, review comments about formatting that a linter should catch. Because "less" is not an accusation, people are willing to put genuinely valued practices here, which is exactly where the honest material is.

Start doing

Practices the team does not have and wants to try. Cap this at one or two per sprint: new habits compete for the same attention, and a team that starts four at once finishes none of them.

Stop doing

Practices with no remaining value. A status report nobody reads, a ceremony that duplicates another, a manual step that has been automated for months. Keep this column reserved for genuine abolition — if the honest answer is "some of it is useful", the card belongs in Less of.

Example board

A nine-person team ran Starfish after a quarter in which delivery was steady but everyone reported feeling stretched.

The two highest-voted cards were connected. "Less of: parallel projects" and "Start doing: written handover" both pointed at the same cause, and the team turned them into one action: no more than one active project per person for the next quarter, with a written handover required whenever that has to be broken. The linter card was done the same afternoon and removed an entire class of review comment.

Timing for a team of 5–10

Facilitator prompts

• Keep doing: "What would you defend if next month got busy?"
• More of: "What did we do once or twice that clearly paid off?" — "What happens only during emergencies that should happen normally?"
• Less of: "What takes more of your week than it is worth?" — "What would you halve tomorrow with no real loss?"
• Start doing: "What did you wish we had when things went wrong?"
• Stop doing: "What do we do that nobody reads, uses or needs?"
• On any card: "Is that a Less of or a Stop doing? Say which, because they lead to different work."

Common mistakes

• Not explaining Less of versus Stop doing. Without the distinction stated up front, the two columns blur and the format loses the advantage it has over Start Stop Continue.
• Overloading Start doing. One or two new habits per sprint survive. Four do not, and the failure teaches the team that retrospective outcomes are optional.
• Treating Keep doing as applause. It is a decision to protect something specific under pressure, not a round of thanks.
• Discussing all five columns evenly. Group, vote, and follow the ranking. A team of nine will produce forty cards and can act on three.
• Skipping the review of last time's actions. Starfish generates commitments every session; without a check, the same cards return and people stop believing the format.
• Running it to understand a bad sprint. It starts from proposals. Discovery needs a different format first.

How to run the Starfish retrospective in QRetro

1. Create a board from the Starfish template — all five columns are already set up.
2. Share the link; participants join in the browser without an account.
3. Read out last sprint's action items and mark what was done before opening the board.
4. Explain the five categories, dwelling on Less of versus Stop doing, then run ten minutes of silent writing.
5. Reveal, group duplicates, and give each person three votes.
6. Work down the ranking and record what changes, who owns it and by when. Stop at three items.