The toolkit · Worked Well, Kinda Worked, Didn't Work
Worked Well, Kinda Worked, Didn't Work retrospective
The blunt three-column retro. Grade the sprint honestly, spend the meeting on the middle column, and leave with one fix that sticks.
30 prompts · 3 columns · free
Worked Well
What actually worked this sprint, no caveats?
10 prompts in this bucket
Kinda Worked
What worked, but only because someone rescued it?
10 prompts in this bucket
Didn't Work
What plainly did not work, however we spin it?
10 prompts in this bucket
Try another framework
Frameworks that pair well with Worked Well, Kinda Worked, Didn't Work.
Start, Stop, Continue
The classic. Three columns, fast to run, hard to argue with.
Mad, Sad, Glad
Emotional check-in. Surfaces the stuff data can't see.
DAKI
Drop, Add, Keep, Improve. Action-first, easy to translate to a backlog.
Or see all 11 retro frameworks.
About the Worked Well, Kinda Worked, Didn't Work retrospective
Worked Well, Kinda Worked, Didn't Work is the retrospective for teams that grade themselves the way they would grade someone else's sprint. Three columns, three verdicts, no room for the vague optimism that turns retros into applause. The middle column is the whole point. Most sprints do not split neatly into wins and failures. They are full of things that shipped late but shipped, processes that held because one person quietly carried them, and meetings that produced half a decision. A Start Stop Continue board flattens all of that into Continue. The Kinda Worked column catches it while it is still cheap to fix, which is why this format tends to find problems a quarter before they would otherwise introduce themselves. Use it when the team's default answer to 'how was the sprint' has been 'fine' one sprint too many.
Want the written version with examples and a copyable board? Read Read the full Worked Well, Kinda Worked, Didn't Work template guide.
Why the middle column matters
Kinda Worked is where process problems live before they become incidents. The deploy that needed a manual step nobody wrote down. The handoff that worked because two people happened to be online at the same time. The estimate that held only because someone worked the weekend. None of these belong in Didn't Work, because the work landed. All of them are one busy sprint away from moving there. Teams that only sort into win and loss keep re-discovering these items at the worst possible moment. Naming them while they still count as wins is the cheapest fix the team will ever get.
How is this different from Start, Stop, Continue?
Start, Stop, Continue is directive. It asks what the team should do differently and assumes everyone already agrees on what happened. Worked Well, Kinda Worked, Didn't Work is evaluative. It grades what happened first and lets the actions fall out of the grading. Run this format when the team does not yet share a picture of the sprint, when one half thinks it went fine and the other half is quietly furious. Run Start, Stop, Continue when the diagnosis is already agreed and you just need commitments.
How long should it take?
Thirty minutes for a two-week sprint. One minute of context, twelve minutes of silent writing, twelve minutes of discussion weighted heavily toward Kinda Worked, and five minutes to vote and commit. If the meeting runs long, it is almost always because Didn't Work turned into a debate about one item. Timebox that item to five minutes and give it an owner instead of a verdict.
Three ways to ruin this retro
First, letting Kinda Worked become the politeness column, where real failures go to avoid hurting feelings. If it did not deliver, it goes in Didn't Work, kindly and specifically. Second, treating Worked Well as a formality to rush past. The team needs to know what to protect, otherwise next quarter's process improvement quietly deletes the thing that was holding everything up. Third, leaving without an owner. A three-column verdict with no owner is a mood board.
Worked Well, Kinda Worked, Didn't Work retro prompts
Every prompt in the Worked Well, Kinda Worked, Didn't Work deck, grouped by column. Shuffle them in the builder above or use the full list here.
Worked Well
- What actually worked this sprint, no caveats?
- What would you defend if someone proposed changing it?
- What went smoother than it had any right to?
- Which decision paid off fastest?
- What did we ship that we are genuinely proud of?
- Where did the process quietly do its job?
- What worked so well we stopped noticing it?
- Which handoff felt effortless this time?
- What would you tell another team to copy from us?
- What deserves credit that will not get it unless we say it now?
Kinda Worked
- What worked, but only because someone rescued it?
- What got done but cost more than it should have?
- Where did we hit the goal but hate the route?
- What worked this time but will not survive a busier sprint?
- What needed a workaround to limp over the line?
- What is running on one person's heroics?
- Which process worked on paper but wobbled in practice?
- What did we finish that nobody wants to do again that way?
- Where did good enough quietly become the standard?
- What would have failed if one thing had gone differently?
Didn't Work
- What plainly did not work, however we spin it?
- What did we plan that never survived contact with the sprint?
- Where did we lose the most time for the least value?
- What broke twice?
- Which meeting produced nothing both times we ran it?
- What did we commit to and quietly abandon?
- Where did the same problem show up for the third sprint running?
- What assumption turned out to be wrong?
- What frustrated the most people this sprint?
- What would you delete from the sprint if you could rerun it?
Frequently asked
- Is Kinda Worked just a nicer way to say Didn't Work?
- No, and policing that line is the facilitator's main job. Kinda Worked is for things that genuinely delivered but cost too much, needed a rescue, or will not survive a busier sprint. If the thing did not deliver, it belongs in Didn't Work. When the team starts using the middle column to soften failures, the format loses the honesty that makes it worth running.
- How many items per column?
- Three to five. The middle column can run longer, and that is usually a good sign, because it means the team is being honest about the gray zone. If Kinda Worked has ten items, dot-vote and discuss the top three. The rest are documented and will resurface if they matter.
- Can we run it async?
- Yes, and the verdicts make async easier than most formats. Open a doc on Monday with the three columns, let people add cards through Wednesday, then meet for fifteen minutes on Thursday to argue about anything that sits between two columns and pick one fix. The between-column arguments are the valuable part, so do not skip the short live session.
- What if everything lands in Worked Well?
- Either you just had a genuinely great sprint, which happens and deserves the recognition, or people do not feel safe grading honestly. Run one round with anonymous cards and compare. If the anonymous board suddenly grows a middle column, the problem was never the format.
- What if the room splits on where an item belongs?
- That argument is the retro. An item half the team calls Worked Well and half calls Kinda Worked means the cost of that work was invisible to the people it did not land on. Let the disagreement run for five minutes. You will learn more from it than from any column total.
When not to use it
Teams mid-conflict or straight after a failed launch. A verdict format hands ammunition to whoever is already assigning blame. Run Mad, Sad, Glad first so feelings get named as feelings, then grade the work a sprint later when the room can hear it.
How to run a Worked Well, Kinda Worked, Didn't Work retro
- 1Block 30 minutes. The format is fast because the columns are verdicts, not open questions.
- 2Recap the sprint in 60 seconds, plain language, no judgement yet.
- 3Show the three prompts and give everyone four minutes of silent writing per column.
- 4Read Worked Well aloud first. It sets the baseline and stops the meeting from opening on a complaint.
- 5Spend most of the discussion in Kinda Worked. That column holds the problems that are still cheap to fix.
- 6Dot-vote across Kinda Worked and Didn't Work, two votes per person. Pick one fix and give it an owner.
- 7Post the board and the single commitment in Slack within the hour.