In this post, we continue a conversation I started in my previous post about large item estimation in backlog grooming.
Another method for estimating backlog items, particularly when there’s a lot of them to estimate, works like this:
- Get your backlog items printed out on cards and get a room with a wide wall space (at least 2m or 6ft, minimum).
- Read each backlog item to the entire team.
- Have the team arrange the items horizontally on a wall in order of size with the easiest items on the right side of the wall and the hardest items on the left side of the wall. Items of similar complexity would be relatively close together on the wall. Items with very different complexity would be relative far apart on the wall. There is NO TALKING allowed during this part of the exercise. The entire effort should take between 10 and 15 minutes.
- Give the team a final opportunity (another 5 minutes) to make adjustments to the ordering. Again, NO TALKING.
- Arrange story point values above the list of backlog items on the wall (use PostIts or cards and tape…). Use the sequence that you normally use with your team (Fibonacci, doubling, or custom).
- Ask the team to group the items around the nearest number that makes the most sense for the backlog items on the wall (now they aren’t estimating individual items, they’re actually estimating groups of items). Give them five minutes, see where they are, discuss any difficulties they may be having, and give them five more minutes. Repeat as needed until finished.
- Allow talking, but now you’ll have to monitor to ensure that the team doesn’t start having technical solution discussions.
- Use a table instead of a wall (things fall off walls more easily than a table).
- If the same backlog item keeps getting moved back and forth on the wall, take it “out of play” until the team is done, then return it when the rest of the items are completed.
Note: I can’t take sole credit for this method. I know that Lowell Lindstrom has talked about a technique called “Affinity Estimation” that is pretty much identical.