KISS Review Framework: Keep, Improve, Start, Stop
The KISS Review Framework helps you turn a look back at recent work into four action buckets (Keep, Improve, Start, Stop) so the room leaves with what continues, what changes, what begins, and what ends.
Framework Card
- Name:
- KISS Review Framework
- Goal:
- Turn a review of recent work into four action buckets: what to keep, improve, start, and stop.
- Flow:
- Keep → Improve → Start → Stop
- Best For:
- Sprint retrospectives; Project post-mortems; Structured 1:1 feedback on habits and actions
Why it matters
Many reviews fail for a boring reason: people spent the hour, and nobody can say what happens next.
"I think we did okay" is an impression. It does not tell a team whether daily standup stays, whether the handoff checklist needs a rewrite, whether to try pairing on the risky tickets, or whether to drop a status meeting that never changes a decision. Projects drift when effort continues and reflection never becomes a keep / change / begin / drop call.
That matters at the end of a sprint, after a delivery, in a quarterly review, and in a 1:1 about habits. KISS is a way to force those four calls. It will not tell you why the sprint slipped. That is a different job.
What it is
This KISS stands for Keep, Improve, Start, Stop. It is a review tool, not the "Keep It Simple Stupid" slogan.
You can treat the four words as buckets. Production teaching also lays them on a 2×2: results on one axis (good / bad) and whether the practice can be sustained on the other.
| Good result | Bad result | |
|---|---|---|
| Sustainable | Keep | Improve |
| Unsustainable | Start | Stop |
- Keep: It works and you can keep doing it. Stay with the method, habit, or idea and build on it.
- Improve: You are still on a viable path, but the result is weak. Change the process or tools; do not throw out the direction.
- Start: Something already helps (or would help) but is not yet a stable practice. Begin it so the good result can last.
- Stop: It fails and you cannot keep it. Drop the method, habit, or behavior.
The matrix is a sorting aid. You do not need an item in every cell. KISS is flexible in *when* you run it (mid-project, at the end, after completion). The four buckets stay the same.
How it works
Run KISS as a sort, not as a tour of feelings.
1. Name the slice of work
One sprint, one launch, one quarter, one habit in a 1:1. A KISS that covers "the team in general" fills with slogans.
2. Collect candidate practices
What did people actually do: meetings, tools, handoffs, tests, communication habits. Prefer items someone could keep or stop next cycle.
3. Sort with the four buckets (and the 2×2 if it helps)
Ask, for each item: was the result good or bad, and is this way of working sustainable? That yields Keep, Improve, Stop, or Start. If two people disagree, the disagreement is usually about evidence of the result, not about the labels.
4. Leave with actions, not a mood
Each bucket should contain things the group can do. Keep items need an owner only if they would otherwise fade. Improve and Start items need a next attempt. Stop items need a date they actually end.
You can use KISS after a GRAI-style look at goals versus results. GRAI finds the gap; KISS files the resulting moves. If the gap is already obvious, skip straight to the buckets.
How it compares
When another lens fits better, these methods do different jobs. They are not interchangeable retro stickers.
Related review lenses
| Framework | What it helps you see | How it differs from KISS |
|---|---|---|
| After-Action Review (AAR) | Intended vs actual, why the gap, what to do next time | Causal learning. Focus on what happened, not who to blame. KISS skips why and sorts actions. |
| 3R Review Framework | Record, Reflect, Refine | A compact cycle, often personal or after a single session. Reflection includes motives. KISS is four action buckets, usually in a team review. |
| PDCA Model | Plan, Do, Check, Act | An improvement loop you run over time. KISS is a classification at a review, not the loop itself. |
| Start-Stop-Continue | Three action lists | No separate Improve. Continue mixes "it works" with "same path, needs a better method." |
| GRAI Review Framework | Gap between goals and results, then insight | Finds what to think about. KISS is how you file the resulting actions. Production copy treats them as complements. |
KISS is the lens when the room needs four action buckets. Other methods help when the question is why, how to run an experiment over time, or how one person records and refines a process.
When to Use This Framework
- Sprint retrospectives: Habits and outcomes from the last increment, sorted into continue / change / begin / drop.
- Project post-mortems: After delivery, capture lessons as follow-up actions rather than a feeling about the project.
- Quarterly reviews: Execution choices next to priorities: what operating rhythm stays, what changes.
- 1:1 feedback: Actions to reinforce or change, instead of a personality sketch.
Example
A concrete example makes the structure easier to reuse when you are under uncertainty.
Example: Sprint retrospective on a payments squad
The squad shipped a payments fix late. Standup still happens every day. The "definition of ready" is ignored. Two engineers paired on the gnarly tickets and those tickets landed clean. A Thursday status meeting produces slides and no decisions.
A KISS-style sort might look like this:
Keep: Daily standup (short, useful). Pairing on the high-risk tickets.
Improve: Definition of ready. The path (ready before build) is right; the checklist is ignored, so results stay bad. Tighten the checklist and refuse work that skips it.
Start: A written "minimum shippable" for payments so late scope cuts are a named set, not a scramble. Pairing helped but is not yet a default on risky work.
Stop: The Thursday status meeting. Unsustainable calendar load and a bad result (no decisions).
Implication: The retro produced four calls. It did not explain the root cause of the delay. If the room still needs *why* the estimate missed, that is After-Action Review work, not more buckets.
Example: 1:1 about calendar habits
A manager and a report review a month of missed writing time.
Keep: Friday afternoon for deep work when it actually happened.
Improve: Calendar holds exist (sustainable idea) but get stolen (bad result). Protect them with a named backup slot instead of abandoning holds.
Start: A 15-minute end-of-day note so unfinished writing has a restart point.
Stop: Accepting "quick" meetings into the Friday hold. That pattern is both unsustainable and producing the bad result.
Production CMS has no example block. These scenes are educational illustrations, not Apply templates.
Takeaway
What KISS can help with
- Turning a review into four explicit action types
- Separating "keep going" from "same direction, better method"
- Making Start and Stop first-class, so the group does not only polish what already exists
- Structuring a 1:1 around habits instead of vague impressions
What KISS cannot replace
- Causal learning: After-Action Review asks what was intended, what happened, why, and what to do next time. KISS does not explain why.
- A personal Record / Reflect / Refine cycle: 3R is a compact learning loop, including motives. KISS sorts actions; it does not ask you to journal the process.
- A Plan-Do-Check-Act loop: PDCA is how you run an improvement experiment over time. KISS is a sort at a review point.
- Start-Stop-Continue: Three buckets. Improve is not a separate call; "continue" mixes Keep and Improve.
- Goal-versus-result diagnosis: GRAI looks at the gap first. KISS files actions once that gap (or the outcome) is clear.
- The Keep It Simple slogan: Brevity advice is a different idea that happens to share letters.
Honest scope: KISS makes the next step explicit. It should not be sold as analysis, and it should not be used as a label for team-development stages.
Frequently asked questions
No. On this page KISS means Keep, Improve, Start, Stop. The simplicity slogan is a different idea that shares the same letters.
Start-Stop-Continue has three lists. KISS splits "continue" into Keep (it works) and Improve (same direction, weak result), and it treats Start as its own bucket, including practices that help but are not yet stable.
AAR reconstructs intended vs actual and asks why. KISS assumes you already know enough about the outcome to classify practices. Use AAR when the room still needs the gap and the cause. Use KISS when the useful output is keep / improve / start / stop.
3R is Record, Reflect, Refine: capture what happened, look at motives and patterns, then adjust the method. It is a learning cycle, often personal or lightweight. KISS is a four-way sort of actions, more often in a team retro or 1:1.
Yes. Production teaching is explicit that the same four buckets work mid-project, at the end, or after completion. The slice of work still needs to be named so the lists stay specific.
Four short lists of practices someone could actually keep, change, begin, or drop in the next cycle. If the output is a mood ("we care about quality") or a theory of team stages, the buckets were not used.