Should We Stop Estimating Hours?
Have you and your team ever caught yourselves trying to convert story points to hours when planning tasks? Today, this is exactly what happened to my team and me. We were sitting around the table planning the tasks for the upcoming sprint with the regular planning poker. When we took the tasks from the backlog and added them to the sprint, we estimated both story points and the time we believed it would take to complete them.
After the planning one of my team mates asked, “Why are we estimating both story points and hours?”. This question lead to a team huddle were we discussed this and we concluded to stop estimating hours and only estimate story points.
The reason to stop estimating hours
- The velocity can be all over the place.
- Story points is easily confused with time (in our case it was hours multiplied with a factor of 3).
- It is hard to accurately estimate hours.
- Two persons on the team might finish a task at different speeds. Moreover, when estimating, the fast person and the slow person will probably disagree on the time it takes to complete the task and therefore has to compromise. Then the task will be either underestimated or overestimated based on the person that works on the task.
- If you tend to be optimistic in your estimation, this might lead to underestimation.
The reason for estimating story points
- Story points can become more accurate and stabilize over time.
- The velocity during a sprint can be pin-pointed.
- It can show uncertainty in the tasks (by using fibonacci).
- The person that executes and completes the task does not matter since the estimate is based on complexity and relative to other tasks.
- Instead of how much time one person needs to finish the tasks, it will show how many story points the team will complete in a given time.
After we concluded by only using story points, we had to quickly reestimate the tasks. I read a blog post a couple of months ago that explained The Silent Grouping Technique. This activity was quite fast and we estimated 30 tasks in around 45 minutes.
This is how we completed the activity:
- Print out all the tasks and make small flash cards with each task.
- Each person takes one task and places it on a whiteboard without saying anything.
- The first person places a task in the middle.
- Then each person places a task either to the left or to the right of the tasks on the whiteboard. Placing it to the right means that it is bigger, and placing it to the left means that it is smaller.
- Continue doing this until all tasks are placed out.
- Then each person goes up, in turn, and moves one task that he/she thinks is placed wrong.
- This is also done without saying anything.
- Continue until no more tasks can be moved.
- If the same tasks are continually moving back and forth, you should stop.
- Agree on estimation sizes
- This can be t-shirt sizes, fibonacci or animal sizes (we chose fibonacci because it shows uncertainty when the tasks are larger).
- Then give each task a size from left to right.
- It is important to continually asking yourselves, is this task twice or three times as big?
The estimated tasks were placed on a cardboard so that it could be used as a answer sheet when we want to estimate new tasks.
After completing the first sprint with the new estimation, we will have a starting benchmark on how many story points we can complete in one sprint. When planning the next sprint we will take in approximately that many story points. Then over time, the benchmark will normalize and become more accurate.