It was a scene you wouldn’t quite expect at a typical company meeting.
There we were: the ReQtest team; cards in hands, clutching them close to our chests and eyeing each other carefully before making our move.
No, we weren’t playing out our fantasies of winning millions at Vegas. What we were doing in fact was using a technique called planning poker to assess how long it will take to complete the task we were discussing at that moment.
Planning poker, also known as Scrum poker, is an easy and effective technique which we often use to estimate the development time needed to create any features we implement. The technique is ideally suited for Agile work, however it will work just as well in conjunction with any other method.
In this article I’ll elaborate on the principles behind planning poker and the reasons why it works.
A helping (poker) hand
The objective of planning poker is basically to force each team member to independently assess their own estimate of the effort required to complete a task and then compare it with the
perception of others. This encourages a collaborative approach in order to reach a consensus among multiple participants on the duration of effort required to finish a task.
When the participants get a set of cards with numbers representing the duration of the task, they choose a card which reflects their estimation of the effort required.
The cards are initially kept private until everybody picks one and only then are the cards turned over for everyone to see. The highest and lowest estimator now have to defend their estimates. If applicable, the rest of the team can complement with their opinions.
Then a second round of estimation take place and the process continues until the values of the cards chosen by the members are close enough.
Estimating the duration of a task individually often yields numbers that are wildly off the charts; however when approached as a group, the consensus that emerges is uncannily accurate. This is popularly known as the ‘wisdom of crowds’ and the same process works in many situations.
A word of caution is in order at this point. It is not possible to estimate time for tasks that are too big. In these situations a rough timeframe would be possible, but nothing more accurate than that. For this reason it is of paramount importance that activities are broken down into the smallest possible sub-tasks before being presented to the team for estimation.
A common question is "how much time do you spend on planning poker?" At ReQtest we use sprints that are two weeks long. Planning of one sprint typically takes three hours, starting with the product owner presenting requirements, and ending with writing stickies and estimating. This is why the poker deck often contains a coffee cup-card. You need to take breaks and keeping the meeting going for three hours without breaks is not advisable. If somebody puts the coffee card on the table it is a sign that you have not planned the meeting perfectly.
Laying the cards on the table
The principles behind planning poker are backed up by various papers that presented empirical evidence about the effectiveness of ‘crowdsourcing’ estimations.
The importance of working only with small tasks is clearly shown in an experiment conducted by Simula research labs in 2006. A group of people were told to estimate the time needed for implementing a requirements document that was presented in two different formats: one copy was much shorter than the other, although both contained the same text.
The people who were shown the bigger document instinctively responded that it would take much longer for its requirements to be implemented than those in the shorter document. This shows that it is important to estimate small tasks one at a time and that larger tasks ought to be split into smaller sub-tasks.
As a follow-up to the same experiment, a group of people had to estimate the time required for implementation of the same document presented either as a brief concise document or in a more complicated format. Again, the document that was more difficult to process was judged to take longer to implement. This highlights the fact that concise and clear writing should be used when formulating requirements. Likewise, when using planning poker, user stories are used to present the relevant information in a simple and easy-to-understand manner.
A card up your sleeve
In many ways planning poker uses the same psychological principles behind ‘groupthink’ but adapts them for a more positive outcome. This is achieved by asking the members to flip their poker cards over at the same time. In this way, any influences by other members is reduced and the estimation shared is a true reflection of how long that person believes the task will take.
Diversity isn’t a problem in planning poker. Often, participants who give longer estimations may be aware of some obstacle that the other members did not consider; for example testing might involve browser testing on many different browsers and versions which may take significant time. Likewise, a participant who gives a very short estimation may be considering a shortcut that the others didn’t think of. This diversity is the basis of the discussion that follows every round of estimation and encourages members to speak out their unique perspectives.
Play it right
If you would like to start using planning poker with your team you don’t need to buy any fancy cards. Just head over to http://planningpoker.com/ and sign up for a free account that will let you use an excellent online version of the planning poker tool.