Every retro I've ever run starts the same way. Screen shared, columns loaded, and a long quiet second where everyone waits for someone else to drop the first card.

So I got curious about what actually lands on those boards. We host a lot of retrospectives on Kollabe, so I pulled the aggregate: 79,868 retros, 986,317 cards. Anonymised, counted, nothing fancy.

The thing that surprised me most: retros have a reputation as a place to air what's broken, and the data says the opposite.

Teams write almost twice as much about what's going right. The "What Went Well" column collects 5.4 cards on average. Its "What we want to improve" neighbour barely clears 3. "Glad" beats "Mad" by more than two to one. Whatever the format, the positive side fills up first.

And the loneliest column in agile? "Stop." At 1.5 cards on average, it's the one teams reach for least. Starting new things is easy. Admitting you should stop doing something, apparently, is not.

The textbook format isn't the popular one. Start, Stop, Continue is the one everyone learns. But the single most-used format is the plainer What Went Well / What Went Wrong / What We Want to Improve, ahead of Start/Stop/Continue and Glad/Sad/Mad. We don't pre-select a format for anyone, so that's a real preference rather than a default nudge.

Friday is retro day. Measured by when a board actually gets filled in (not when it was created the day before), Friday is the most popular day to run a retro, just ahead of Tuesday. Teams reflect at the end of the sprint and the end of the week.

Action items are where good intentions go to rest. A typical retro produces about three of them. Only around one in eight ever gets marked done. I'd love to say my own team's board looks better than that. It does not.

There's more in the full breakdown, with charts for formats, columns, timing, kudos and reactions: Retrospective Statistics 2026.

A rule of thumb I've started using: if your retros only ever fill the "what's broken" column, you're probably running a complaints session, not a retrospective. The good stuff is a signal too. It tells you what to protect when the next reorg or deadline comes for it.

The honest caveat: this is data from teams who chose Kollabe, so it skews toward teams who like running structured retros in the first place. Your mileage will vary. But a million cards is a lot of cards, and the positivity gap holds across every format we looked at.

If you want to run one this week, Kollabe retros are free to start. Or grab whatever board you already use, add a real "Stop" column, and watch how empty it stays.