DAKI (Drop / Add / Keep / Improve)
Improve is where the soft cards go to die. Force a name and a deadline on each one or drop the column.
Four columns — Drop, Add, Keep, Improve. The differentiator is the split between Keep and Improve: it lets a team name something that's working AND needs refinement. DAKI has no agreed origin and no canonical author; it's an emergent acronym, not a methodology.
When to use
After a sprint where the team hit the goal but the path was ugly. After a delivery that worked but the process around it didn't. Skip it for a clean sprint — SSC is faster. Skip it if the team treats Improve as a wishlist; without owners on every Improve card, the column collects nice ideas nobody touches.
How it runs
Four columns
Drop, Add, Keep, Improve. Read them in that order — Drop forces a real decision, Add is energising, Keep is reassurance, Improve is the work.
Silent write, six minutes
Standard silent write. Senior person writes last. The Improve column will be empty if the team starts with Keep — write all four columns at once.
Cluster, then triage Improve early
Group duplicates. Read Drop and Add briefly, then go straight to Improve. Keep is the shortest discussion, not the longest.
Force ownership on every Improve card
Each Improve card needs a name and a deadline before it leaves the meeting. 'We should improve our standups' with no owner is a card that returns next sprint unchanged. If you can't find an owner, drop the card.
One Drop, one Improve
Two commitments. Drop is the action; Improve is the refinement. Don't try to do three.
Why it works
Most four-column retros conflate 'this works' with 'don't change it.' DAKI separates them — Keep is for things that work and shouldn't be touched; Improve is for things that work and should be refined. The split surfaces a category SSC and KALM don't have a column for: the half-broken thing the team has been tolerating because it isn't broken enough to Stop.
Variations
- Drop the Improve column if it keeps filling with unowned wishes — at that point you're running SSC with a Drop column instead of Stop, which is fine.
- Run it after a tooling change or a process change — DAKI is the format that handles 'the new tool works but it's awkward' as its own category.
- Pair it with a 15-minute timeboxed Improve discussion at the end of the meeting — protects the column from being the part everyone skips.
Facilitator notes
The most common failure mode: Improve becomes the column the team uses to avoid Drop. 'Drop the daily sync' is hard; 'improve the daily sync' is easy. When you see an Improve card that should be a Drop, ask the writer 'is this actually fine if we change it, or are we trying not to say Stop?' That's the format earning its keep.
Pitfalls
- Improve as a wishlist. Every Improve card needs a name and a deadline or it's filler.
- Treating Drop and Improve as the same column. Drop ends a thing; Improve refines it. Push back when they bleed.
- Pretending DAKI has a lineage. It doesn't — no Norman Kerth, no Luke Hohmann. It's an emergent acronym; say so when someone asks where it came from.
- Running it as the default sprint format. SSC is faster and pushes harder for action.
Remote tips
Sort the Improve column by votes during the live discussion — visible ranking stops the loudest voice from anchoring on their pet card. Tag each Improve card with an @ mention before closing the meeting; the @ is what survives the call.
Example outputs
- Drop: the design review queue as a separate step. Inline review on the ticket instead.
- Add: a sprint-end demo to the support team. They flag bugs we miss.
- Keep: the Tuesday sync at 30 minutes. Don't make it longer.
- Improve: standup. Owner: Priya. Deadline: end of next sprint. Three-sentence updates, not three-paragraph monologues.
FAQ
- Should I run DAKI or Start/Stop/Continue?
- DAKI when the team's wins and improvement areas overlap — when something's working AND needs refining and you want a column for that. SSC when the categories are clean: this works, this doesn't, do more of this. If you're not sure, run SSC — it's faster.
- How is DAKI different from KALM?
- Both are four-column SSC variants. KALM softens Stop into Less and adds More. DAKI keeps a hard 'Drop' and adds Improve. KALM is for teams that need 'Stop' softened; DAKI is for teams that want a column for half-working things they'd like to keep. Different problems, different formats.