The Science Behind Planning Poker: Understanding the Theory That Makes It Work

Planning Poker has become the de facto estimation technique in agile teams. Like many agile practices, it elegantly packages theoretical concepts into a practical, easy-to-follow process. This post explores the hidden foundations of Planning Poker by combining three powerful ideas: measurement theory, research on cognitive anchoring effects, and the Delphi estimation technique. Understanding these underlying principles helps teams move beyond just following the process to truly mastering the art of collaborative estimation.

Overview of planning poker process

Just to frame the conversation. Planning poker is a shared estimation technique that many agile teams use to estimate the complexity of a user story.

From a birds-eye view, the planning poker process goes like this:

Set up:

Each team member has a deck of cards with “points”. The same deck of cards is given to each of the team members

Process:

1. A user story is read.
2. Each team member puts, face down, a card on the table. The number in that card represents the team members’s estimation.
3. All cards are flipped at the same time.
4. Team members presenting extreme points, explain their reasoning.
5. The process is repeated until the team is happy with the dispersion and a number of points can be assigned to the user story.

The Power of Shared Cards

At first glance, having team members simultaneously reveal their estimate cards might seem like a simple way to make estimation more engaging. However, this practice is rooted in research on cognitive biases, particularly what psychologists call the “anchoring effect.” When estimates are shared sequentially, team members unconsciously anchor their thinking to the first numbers they hear, especially from senior team members or leaders.

The simultaneous reveal in Planning Poker isn’t just about making estimation fun – it’s a deliberate counter to this cognitive bias. When teams eliminate anchoring effects, they tap into genuine collective intelligence, often outperforming even their most experienced individual estimators.

Why Peer Estimation Matters

The peer estimation aspect of Planning Poker draws directly from the Delphi technique, a structured communication method developed in the 1950s. The benefit of the delphi method lies in its ability to harness collective intelligence while minimizing biases and ensuring that each expert’s opinion is considered equally. I won’t get into the details, yet the core of the process lies in the next three properties:

1. Independent Thinking: Each team member forms their own opinion before being influenced by others.

2. Structured Discussion: When estimates differ, teams explore the reasoning behind different perspectives

3. Iterative Refinement: Multiple rounds allow teams to converge on realistic estimates

The practical software measurement data consistently shows that the Delphi-based approach delivers:

– 20-30% more accurate estimates compared to individual expert estimations,

– Better identification of potential risks and complexities,

– Improved team understanding of the work scope.

Planning Poker has distilled its core principles into actionable steps that have become popular in agile teams.

The Role of Historical Data

I like to emphasise the importance of calibration through historical data. The best practice is that the estimation dimension in planning poker should not be “effort”. It should be some other dimension (like complexity, or size). A common mistake from teams that start with estimating planning poker is that they anchor the complexity until to effort (“How much is 8? Ans “think 8 points equals one day”). This anchoring is then very difficult to untangle. 

Maintain a simple record of:

– Common sources estimation deviation: What are the most common discussion topics when dealing with estimation extremes? Take note of these, and take them to the retrospectives. They could be a source of training and technical improvement.

– Estimation vs Effort. Is effort estimation still a requirement within your organisation? Ask for patience and keep track of the relationship between your complexity/size dimension and effort. Eventually, you’ll be able to discover a relationship between the two and estimate effort by using Complexity/size as a proxy.

Best Practices for Effective Planning Poker

Before the Session

Try these two things to prepare for effective Planning Poker sessions.

– INVEST in your user stories. Many issues come from ill-defined stories or stories that are not comparable with estimation sets. Review your story writing practice (If you need help ask me about the Workshop “best practices for writing User Stories”).

– Set clear estimation criteria. Make sure the whole team understand the dimension that is being estimated (complexity, size, or uncertainty).

During Estimation

There are also two keys to effective sessions.

1. Allow silent reading and consideration time. Teams make a lot of decisions during planning sessions and take on considerable risks. It is important to give adequate time for consideration of each story. Don’t rush to get a number and give “a reasonable” quiet time for each story estimate to be considered.

2. When estimates differ: Ask for explanations and focus discussion on understanding different perspectives, be mindful that you are not looking for technical solutions for the user story. You are looking for hidden criteria and risks that might have not been evident to the whole team.

Parting thoughts

Remember: the goal isn’t perfect estimates, but rather consistent relative sizing that helps teams plan effectively and deliver value predictably. The conversations that happen during Planning Poker often prove more valuable than the numbers themselves.