Is KALM the same as Start, Stop, Continue?

It looks similar but works differently. Start, Stop, Continue is binary: a thing is either added, removed, or kept. KALM is gradient: a thing can be made smaller or larger without being killed or invented. That nuance matters more for mature teams making small changes than for new teams making big ones.

Can we use KALM after a hard sprint?

Probably not. KALM assumes the team has the bandwidth to think about adjustment rather than recovery. After a difficult sprint, run Mad, Sad, Glad or 4Ls first, then come back to KALM when the team is in steady state again.

How many items per column?

Three to five per column. KALM has four columns, so a maxed-out retro is twenty items, which is too many to discuss. Dot-vote down to one or two per column before discussion.

What's the difference between Add and More?

Add is for things that don't yet exist. More is for things that exist but should happen at higher volume. 'Add a weekly architecture review' is an Add. 'More pairing on tricky tickets' is a More. They're easy to confuse and the distinction matters less than the discussion they produce.

How often should we run it?

Every three or four sprints when the team is in steady state. KALM is not a default format, it's a periodic check-in for fine-tuning. Run it too often and the team will start producing tiny changes for the sake of producing something.

The toolkit · KALM

KALM retrospective

Keep, Add, Less, More. The dial-tuning retro for teams that don't need anything ripped out, just adjusted.

40 prompts · 4 columns · free

🔒

Keep

What should we protect exactly as it is?

10 prompts in this bucket

➕

Add

What's missing from how we work?

10 prompts in this bucket

🔽

Less

What should we do less of?

10 prompts in this bucket

🔼

What should we do more of?

10 prompts in this bucket

Try another framework

Frameworks that pair well with KALM.

Start, Stop, Continue

The classic. Three columns, fast to run, hard to argue with.

DAKI

Drop, Add, Keep, Improve. Action-first, easy to translate to a backlog.

Rose, Bud, Thorn

What's blooming, what's growing, what's prickly. Garden metaphor, real insight.

Or see all 10 retro frameworks.

About the KALM retrospective

KALM (Keep, Add, Less, More) is the retrospective format for the team that is mostly fine. Not the team that's burning down. Not the team that just shipped something hard. The team that is, on most metrics, doing well, and wants to make small adjustments rather than big course corrections. Where Start, Stop, Continue is binary and DAKI is action-first, KALM gives you a dial. Less of this, more of that. Add this small thing, keep that big thing. The Less and More columns in particular are doing real work: they let the team name things that are halfway between right and wrong, where the answer isn't 'kill it' but 'turn it down'. Use KALM when the team has been together long enough to talk in gradients, and when the changes you need are adjustments rather than overhauls.

How long does a KALM retro take?

Forty minutes for a two-week sprint. Four columns means more discussion than Start, Stop, Continue but less weight than 4Ls. For a one-week sprint, twenty-five minutes is enough. For a quarterly, ninety. The Keep column is the one teams skip when running short on time, which defeats the format's purpose: KALM works because the team aligns on what's protected before naming what to change.

When KALM beats Start, Stop, Continue

When 'Stop' would be too harsh. Start, Stop, Continue forces every item into binary buckets, which works for clear changes and overcorrects for fine-tuning. KALM lets the team say 'we should do less of this without killing it'. That nuance matters when the team is mature enough to know that not everything broken needs to be removed, sometimes it just needs to be smaller.

When KALM beats DAKI

DAKI (Drop, Add, Keep, Improve) is action-shaped: every item implies a thing to do. KALM is more reflective. Use DAKI when you want to translate the retro into backlog items. Use KALM when you want to talk about gradients of practice rather than discrete actions. Both are good rotation partners with each other.

Common mistakes with KALM

First: empty Less and More columns. Teams default to Keep and Add because both feel positive. The Less column in particular is the one that produces the highest-leverage changes (a small reduction in meeting frequency, a tighter scope on a recurring practice) and gets undervalued. Second: treating Keep as filler. The Keep column is where the team protects the rituals that are quietly working. It needs equal time. Third: skipping the dot-vote. Without it, KALM produces a long list and no commitments.

Frequently asked

Is KALM the same as Start, Stop, Continue?: It looks similar but works differently. Start, Stop, Continue is binary: a thing is either added, removed, or kept. KALM is gradient: a thing can be made smaller or larger without being killed or invented. That nuance matters more for mature teams making small changes than for new teams making big ones.
Can we use KALM after a hard sprint?: Probably not. KALM assumes the team has the bandwidth to think about adjustment rather than recovery. After a difficult sprint, run Mad, Sad, Glad or 4Ls first, then come back to KALM when the team is in steady state again.
How many items per column?: Three to five per column. KALM has four columns, so a maxed-out retro is twenty items, which is too many to discuss. Dot-vote down to one or two per column before discussion.
What's the difference between Add and More?: Add is for things that don't yet exist. More is for things that exist but should happen at higher volume. 'Add a weekly architecture review' is an Add. 'More pairing on tricky tickets' is a More. They're easy to confuse and the distinction matters less than the discussion they produce.
How often should we run it?: Every three or four sprints when the team is in steady state. KALM is not a default format, it's a periodic check-in for fine-tuning. Run it too often and the team will start producing tiny changes for the sake of producing something.

When not to use it

Brand-new teams or teams in a difficult patch. KALM rewards a team that has settled enough to think about gradients rather than absolutes. For new teams, run Start, Stop, Continue. For teams in pain, run Mad, Sad, Glad.

How to run a KALM retro

1Block 40 minutes. KALM has four columns and rewards depth over speed.
2Open with a 60-second sprint recap. KALM works best when the team is in a calm-enough state to think about adjustment, not survival, so name that state explicitly.
3Show all four prompts. Five minutes of silent writing per category.
4Discuss in order: Keep, Add, Less, More. Keep first sets the floor (what's not for changing), Add introduces options, Less and More are the dial-tuning.
5Vote on Add, Less, and More with two dot-votes per person. Pick one item per category that the team will commit to.
6Close with a verbal Keep round. Each person names one Keep, briefly. The exercise reminds the team what they're protecting before the changes land.
7Send the recap in Slack within an hour, or the Adds and Lesses won't translate to anything in the next sprint.