Almost everything for us is now a pomodoro . Some time ago we replaced the per-iteration planning game with on-demand planning pomodoro s and the end-of-iteration retrospective with a pomodoro retrospective.

Every Wednesday , the first day of our weekly iteration , we kick off with a 25-minute planning pomodoro to give us just enough stories to make a start. We've been experimenting with a modified fishbowl format to stimulate the right level of technical discussion and keep energy levels surging. When the board gets short of stories (and the team is in danger of stalling) we run another planning pomodoro to top up the 'waiting' column. We keep doing this until the showcase on Tuesday but we're careful not to end with loads of stories still in progress.

Our estimation is simply: "Is this story less than 2 days?" and the team shows thumbs up or down. If it's not, the story is split there and then. A pre-planning pomodoro run in advance of Wednesday, looking ahead into the next iteration, helps to size stories appropriately and get the acceptance criteria on the back of the cards. Our velocity is the average-to-date, over all the iterations, of the number of cards that made it to done. This gives us a steadier velocity than summing the estimates of the stories that made it to done in that iteration. We're not really using the velocity for planning purposes though. We use it to calculate a per-story cost, in £, derived from what was delivered in the iteration and the overall capacity cost for the iteration. From this we can then calculate the cost of any inventory and outstanding technical debt. It's sobering to see these things in money terms. I'll blog separately about the simple profit-and-loss sheet we use based on lean accounting and the metrics we watch.

The pomodoro retrospective is conducted standing up. This keeps things moving and keeps people focused. 25 minutes doesn't provide a lot of scope for variation of activities but we can easily cover our standard format: brainstorming - affinity mapping - dot voting - and agree one action that will improve things. I expect this will eventually get boring so I'm thinking of ways to do pomodoro 'lets improve this, here-and-now' sessions that are triggered by a problem just encountered. I guess these are similar to timeouts. Ultimately, the challenge I've set myself is to focus these continuous improvement pomodoros on making an improvement that is not borne out of (and therefore constrained by) solving a specific problem. They just focus on making an improvement to get better.

We're also experimenting with a new board layout that helps us integrate iterative collaboration with designers. There's some serious thinking to be done here and, to be honest, we could do with a fresh project to try it out on.