When improvement options need to be sorted more finely than just start and stop, it brings observations into clear kinds of action. This makes visible what should be done more, less, newly, or continued.
Agile Facilitation method
Starfish Retrospective
Shows which practices should be preserved, increased, reduced, stopped, or tried anew.
The Starfish Retrospective distinguishes five changes to practices: Keep Doing, Less Of, More Of, Stop Doing and Start Doing. Preserving, adjusting intensity, ending and trying something new remain distinct decisions.
Which practices should we preserve, increase, reduce, stop or start trying?
- Participants
- 4-12 participants
- Time
- 45–75 minutes; later experiment review

Five directions
Five directions of change
Set the period and explain Keep, Less, More, Stop and Start using a practice.
Practice list
Collect practices silently
Independently record concrete practices and place them in the desired change category.
Disagreements
Understand disagreements
Clarify related cards and explicitly discuss conflicting change requests.
Adjustment
Test a defined adjustment
Select a feasible practice change with specific extent and expected effect.
Test a selected change in the next work period; at review decide its intensity or continuation using observed effects.
A good fit when
For teams wanting nuanced changes to working practices rather than merely collecting positives and negatives.
Ideal for
- Continuous Improvement
- Mature Teams
- Practice Tuning
Not good for
- Very new teams without shared experience
- Acute conflict resolution
Outputs

Five practice directions
Five differentiated practice lists and an agreed experiment with baseline, intended intensity, owner and review.

Practice experiment with review
The practice experiment records an owner, duration, later review signal, and explicit protection for Keep practices.
Starfish Retrospective· worksheet
Fictional example
The Starfish Retrospective distinguishes five changes to practices: Keep Doing, Less Of, More Of, Stop Doing and Start Doing. Preserving, adjusting intensity, ending and trying something new remain distinct decisions.
Keep Doing
Which practice should remain as it is, and why?
Fictional: keep the Friday demo with real usage cases; it clarifies misunderstandings.
Less Of
Which existing practice should become less frequent or smaller?
Limit status meetings to 15 rather than 30 minutes.
More Of
Which helpful existing practice should become more frequent or intensive?
Increase pairing on unfamiliar modules from once to twice weekly.
Stop Doing
Which practice should stop and which necessary function must still be covered?
Stop the duplicate email status report; the current board remains the information source.
Start Doing
Which currently absent practice do we want to try?
Start a weekly 20-minute review of support patterns.
Selected experiment
Which single change is tested for how long, by whom and using which effect?
Two weeks of shorter status meetings, owner Lea; observe duration and subsequent clarifications. Then decide whether to continue.
All example details are fictional and illustrate the application.
First step
Define period, team boundary and previous experiments; explain the five categories and request concrete practices.
Deep dive
It distributes feedback across five response patterns and produces concrete change options from it. The additional gradation between More, Less, and Keep prevents sensible routines from appearing only as continue or end.
Starfish works because it distinguishes behavior more finely than a pure continue-or-stop logic. Good application uses the middle ground between more of this, less of this, keep, start, and stop to make nuances in everyday team life visible. This gradation in particular helps avoid overreactions and dose change so it stays realistically connectable in the next cycle.
Introduce the five areas briefly using real behavioral examples, so entries don't feel arbitrary. The work is most productive where a point wavers between two fields and thereby demands a finer decision. Close with a few deliberately graded changes rather than a long list of measures.
The method guide walks execution through 4 phases, timing, 4 pitfalls, and clear stop criteria.
Open method guideYour perspective
Does this method help with your work?
Your feedback helps shape the next improvements to methodatlas.
Sources and topics
Starfish Retrospective: WorksheetDefine period, team boundary and previous experiments; explain the five categories and request concrete practices.markdown
starfish-retrospective-working-template.md
Define period, team boundary and previous experiments; explain the five categories and request concrete practices.
Starfish Retrospective: Worksheet
Question
To be confirmed
Desired outcome
To be confirmed
Scope
Complete retrospective Time: 45–75 minutes; later experiment review
Complete this worksheet on paper or in your own document during the work.
Preparation for this scope
Method setup
- Materials: Five fields labelled Keep Doing, Less Of, More Of, Stop Doing, Start Doing; cards and experiment log.
- Roles: Team members with different perspectives and a facilitator.
- Advance information: Define period, team boundary and previous experiments; explain the five categories and request concrete practices.
- Overall time needed: 45–75 minutes; later experiment review
- Setup: Lay out five equal practice fields; provide a separate area for the change selected later.
Prepare for the work steps
Five directions: Five directions of change
- Retrospective period
- Practices
Practice list: Collect practices silently
- Category logic
- Own experiences
Disagreements: Understand disagreements
- Practice lists
- Concrete examples
Adjustment: Test a defined adjustment
- Practice changes
- Capacity
Starfish Retrospective· worksheet
Keep Doing
Which practice should remain as it is, and why?
...
Less Of
Which existing practice should become less frequent or smaller?
...
More Of
Which helpful existing practice should become more frequent or intensive?
...
Stop Doing
Which practice should stop and which necessary function must still be covered?
...
Start Doing
Which currently absent practice do we want to try?
...
Selected experiment
Which single change is tested for how long, by whom and using which effect?
...
Answer the prompts with your own information. Explicitly mark unknowns and assumptions.
Patrick Kua: The Retrospective Starfish
Five directions: Five directions of change
Expected artifact: Shared category logic
Are keeping and increasing clearly distinguished?
Entry:
...
- Are keeping and increasing clearly distinguished?
Practice list: Collect practices silently
Expected artifact: Five practice lists
Does each card name a practice and desired change?
Entry:
...
- Does each card name a practice and desired change?
Disagreements: Understand disagreements
Expected artifact: Reasoned practice changes
Is context understood for conflicting cards?
Entry:
...
- Is context understood for conflicting cards?
Adjustment: Test a defined adjustment
Expected artifact: Practice experiment with review
Are baseline and new intensity described measurably?
Entry:
...
- Are baseline and new intensity described measurably?
Open questions and next steps
...
Method guide: https://methodatlas.meierhoff-systems.de/en/methods/starfish-retrospective/run-sheet
Alternatives and neighbors
Fitting alternatives with decision guidance, followed by other related methods.

Instead of Starfish Retrospective, when you want to organize team work, backlog, or retrospectives in a lightweight work format.

Makes a team's winds, anchors, risks, and target picture visible and turns the review into prioritized improvements.

Sorts review, learning points, and friction into Liked, Learned, Lacked, and Longed For for concrete improvements.

Collects emotional signals from the team, clusters them into themes, and makes real conversation topics visible.

Makes patterns from a sprint visible and derives concrete improvements for the next work cycle.

Compares stories relatively, surfaces knowledge gaps, and produces a shared estimate with traceable assumptions.
Constraints and origin
Profile
- Method form
- Facilitation method
- Participants
- 4-12
- Time
- 45-90 min
- Maturity
- Established
- Origin
- Praxisgemeinschaft

