Using "Planning Poker" to Avoid Bias in Estimation

When estimating the effort of a project, the planning fallacy often causes a build-up of process overrun. Put simply, people always underestimate the amount of time they will take to complete work. Add to this the natural inclination to claim a better production standard than your lateral colleagues in front of top brass, and you have a recipe for a problem.

How does the planning fallacy show itself?

Fallacies usually occur because of bias, and the software dev industry is no exception.

When managers are given the decision making power to estimate schedules, (s)he is usually going to underestimate ground level work for two reasons: 1. There is definite incentive for management to make underlings work fast, and 2. People who are not on the ground level usually cannot relate to the nuances that slow down ground level work.

When individuals are given their own scheduling powers, everyone starts playing politics. No one is going to estimate a week when everyone else is saying three days. Those who do appear inefficient; however, less time usually means lower quality and backend excuses.

What is planning poker?

Planning poker is a literal deck of cards with numbers on them. Each person involved in the project chooses a card to estimate the amount of time that a specific task will take. Everyone shows their card at the same time, and discussion is had about the differences in the estimates. In this way, everyone can explain his or her perspective without feeling the need to outproduce the guy in the next chair.

In group planning poker sessions, high and low estimates tend to cancel each other out. People without a direct responsibility for the task can also offer a completely unbiased third party opinion. This is especially helpful when people with past experience performing the task at hand weigh in based on their knowledge, without the need to understate their own schedule because they feel their job is on the line.

The Politics of Planning Poker

Planning poker is meant to reduce bias as much as possible; however, there are always office politics to deal with. Here are a few rules to ensure the best round of poker with your colleagues.

• Play the game in a room with plenty of space, around a circular table. This lets people look each other in the eye, and the hierarchy of the company does not unduly influence ground level workers.
• Create a baseline from the easiest tasks. Bias is much easier to reduce when you have a standard from which to set the rest of your processes. The task with the least amount of effort assigned to it is usually the best baseline. It will ground the estimates of more complex tasks, making them more realistic.
• Start the poker session with high priority items. You want to baseline mission critical processes to make sure they get done.
• Avoid anchor bias. An "anchor" is a biased estimate that is based on irrelevant outside information. For instance, a management level employee may yell out an estimate before the planning poker game has been played for a particular task. People will naturally base their estimates during the game on the unsubstantiated anchor that was blurted out ahead of time. Do not allow anyone, no matter their rank in the company, the privilege of speaking before the game starts.

The planning fallacy often overlooks software errors that will pop up because of an incompatible library or an unexpected operating system upgrade. You can definitely reduce your bias by using vetted software error reduction packages like Airbrake. Use Planning Poker and the right supplemental tools to create the most effective process for your company.