Planning poker (a mixture of expert opinion, analogy and fun) is the best way for all members of an Agile team to reliably estimate a user story or theme. As an example of how it works each estimator is given a deck of cards reading 0, 1, 2, 3, 5, 8, 13, 20, 40 and 100. A moderator then reads the story or theme to be estimated; brief time is spent in which questions are asked by members and answered by the Product Owner; each estimator then makes a choice and all simultaneously turn them over to be seen.
Chances are the estimates will vary widely, this is good as the debate which follows will lead to a convergence generally by the third round of voting.
According to Mike Cohn in <i>Agile Estimating and Planning</i> a number of points are worth remembering:
• Time is not wasted in preliminary design discussion (perhaps use a two minute egg timer, and don’t turn it over more than twice!).
• If many items have to be considered, think about splitting the team into smaller teams, having first involved them all in a joint planning poker session to establish a baseline.
• Planning poker should be played throughout a project’s life cycle and falls into two distinct periods. The first is before the project begins and a large number of items need to be assessed. This could be quite a few hours long. The second period will occur at the end of each iteration; these will typically be a lot shorter.
• Planning poker works because it brings together all the experts of a cross-functional team – and they are the most competent people to solve the task; lively debate in which estimators are asked by their peers to justify their estimates improves accuracy; and general discussion averages individual estimates, which leads to better results.
We are using “Planning Poker” in all our projects. It is a very very intuitive way to size a project.
It is fun and it works.