January 15, 2026 · By Raksit Nongbua
How Planning Poker Works: A Complete Guide for Agile Teams
Planning poker is one of the most widely adopted estimation techniques in agile software development. On the surface it looks like a simple card game, but its design encodes decades of research on group decision-making, cognitive bias, and the psychology of uncertainty. If you have ever sat through an estimation meeting that turned into a two-hour argument or produced numbers nobody believed, planning poker is the antidote. This guide walks through what it is, where it came from, how to run a session, and what to watch out for.
The Origins of Planning Poker
Planning poker was introduced by James Grenning in 2002 in an article titled Planning Poker or How to Avoid Analysis Paralysis while Release Planning. At the time, Grenning was looking for a way to make release planning faster and more accurate without devolving into the familiar pattern of one or two dominant voices setting all the estimates.
The technique gained wide recognition when Mike Cohn included it in his influential 2005 book Agile Estimating and Planning. Cohn refined the practice, popularised the use of the Fibonacci-inspired card deck, and helped codify planning poker as a standard tool in the Scrum practitioner's toolkit. Today it is used by software teams worldwide, from two-person startups to enterprise engineering organisations with hundreds of developers.
The name itself is a deliberate nod to poker. Both games involve hidden information revealed simultaneously — and that simultaneous reveal is the single most important design choice in the whole technique, as we will explore below.
How a Planning Poker Session Works Step by Step
Running a planning poker session is straightforward once you understand the rhythm. Here is a detailed breakdown of each step.
Gather the team and the backlog
The product owner, scrum master, and all development team members join the session. Everyone needs a set of estimation cards — either physical cards or a digital tool like Corgi Planning Poker. The product owner comes prepared with the user stories that need to be estimated, ideally with acceptance criteria already written.
Read and discuss the story
The product owner reads out the user story and answers clarifying questions. This phase is crucial. Estimates are only as accurate as the team's shared understanding of the work. Good questions at this stage include: What does "done" look like? Are there known technical risks? Does this touch any legacy systems? The goal is not to exhaustively design a solution, but to reach a common mental model of the scope.
Each estimator privately selects a card
Once discussion winds down, every team member independently selects the card they think best represents the relative effort for the story. Cards stay face-down (or hidden in a digital tool). No one announces their estimate yet. This private selection step is where individual judgment forms, free from the influence of others.
Reveal simultaneously
All estimators flip their cards at the same time. This is the poker moment. When all estimates agree or are close, the team picks a value and moves on quickly. When estimates diverge significantly — say, a 2 and a 13 on the same story — those outliers hold the real information. The person who estimated highest and the person who estimated lowest are asked to explain their reasoning.
Discuss outliers and re-estimate
After hearing the outlier rationale, the team discusses briefly. The high estimator may have spotted a hidden dependency. The low estimator may have a shortcut the others hadn't considered. These conversations often surface assumptions that were never made explicit. After discussion, the team votes again. This cycle repeats until consensus is reached, typically within two or three rounds.
Record the estimate and move on
Once consensus is reached, record the story point value against the backlog item and advance to the next story. A well-run session can estimate 10–15 stories per hour. If a single story consumes more than 10 minutes, it is usually a sign the story needs to be split or the acceptance criteria need more work — not a reason to keep estimating.
Why Simultaneous Card Reveal Matters
The simultaneous reveal is not a gimmick. It directly addresses two well-documented cognitive biases that plague sequential estimation: anchoring and groupthink.
Anchoring bias Anchoring bias occurs when the first number stated in a negotiation or discussion acts as a psychological anchor that pulls all subsequent estimates toward it. Research by Tversky and Kahneman in 1974 demonstrated that people adjust insufficiently from an initial anchor even when they know it is arbitrary. In a traditional meeting, the first person to say "I think this is a 5" effectively sets the range of the whole conversation. Planning poker eliminates this by ensuring no estimate is revealed until all estimates are formed.
Groupthink Groupthink is the tendency for teams to converge on a shared view in order to maintain harmony and avoid conflict, often at the expense of critical thinking. In estimation, this means junior developers defer to the tech lead, or everyone adjusts their view once they see the project manager nodding at a particular number. Simultaneous reveal forces every team member to commit to an independent judgment before seeing others' views, producing genuinely diverse estimates when the story has genuine ambiguity.
The result is that planning poker surfaces disagreement that would otherwise be suppressed. That disagreement is valuable — it reveals different mental models of the work, which when reconciled, lead to better estimates and fewer late-sprint surprises.
Choosing the Right Card Deck
The card deck you use shapes how your team thinks about estimation. There is no single right answer, but here are the most common options and when each works best.
Fibonacci Fibonacci sequence (1, 2, 3, 5, 8, 13, 21) is the most popular choice. The gaps between values grow as numbers increase, which reflects the reality that large stories carry more uncertainty and that false precision in high estimates is misleading. Estimating a story as 19 rather than 21 implies a level of accuracy that does not exist.
T-shirt sizes T-shirt sizes (XS, S, M, L, XL) are useful for early discovery or roadmap-level estimation where a team is trying to size epics rather than individual stories. They remove the temptation to equate numbers with hours, which is especially helpful when stakeholders are in the room.
Powers of 2 Powers of 2 (1, 2, 4, 8, 16, 32) provide even wider gaps than Fibonacci and work well for teams doing rough capacity planning across quarters.
Beyond number cards, most decks include a ? card for "I don't have enough information to estimate," an infinity card for "this story is too large to estimate and needs to be split," and a coffee cup card to call for a break. These non-numeric cards carry as much information as any number and should be treated seriously rather than jokingly.
Common Planning Poker Mistakes (and How to Avoid Them)
1. Estimating tasks instead of user stories
Story points measure the complexity and effort of delivering a slice of user value, not a technical subtask. When teams break work down to the task level before estimating, they lose the ability to use velocity meaningfully. Estimate stories at the level where a user can recognise the value delivered.
2. Letting the product owner dominate the discussion
The product owner answers questions and clarifies requirements but should not estimate. Their job is to define what, not how much. When product owners express opinions about effort ("this should be easy"), they inadvertently anchor the team's estimates.
3. Equating story points with hours
This is the most common mistake and arguably the most damaging. Story points are relative, not absolute. A 5-point story is not "five hours of work." The moment a team starts thinking in hours, the estimate becomes a commitment rather than a forecast, and the psychological pressure to hit estimates overrides the cognitive benefits of relative sizing.
4. Skipping stories after the first vote converges
When the first round of estimates all agree, teams are tempted to skip discussion entirely. Occasionally, unanimous agreement simply means everyone is equally uncertain. A quick ten-second gut-check — "does everyone feel genuinely confident about this?" — costs almost nothing and catches the stories where false consensus is masking hidden risk.
Planning Poker Online vs In-Person
Planning poker was originally designed for co-located teams with physical card decks. Today, the majority of agile teams are distributed or hybrid, and online planning poker tools have become the default. Digital tools offer several advantages over physical cards that are worth naming explicitly.
The simultaneous reveal is easier to enforce online. With physical cards, someone can peek or reveal early. A well-designed online tool locks all estimates until every participant has voted and then reveals them all at once.
Statistics are surfaced automatically. After a reveal, a good tool shows the average, the spread, and who voted what — giving the facilitator immediate data about where disagreement lies without manual counting.
Remote participants are first-class citizens. In a hybrid meeting where some people are in a room and some are on video, physical cards exclude remote participants. An online tool treats every participant identically regardless of location.
History is preserved. Online sessions keep a record of estimates, which is invaluable for retrospectives and for comparing estimated vs actual velocity over time.
Corgi shows the vote average, spread, and every individual vote the moment the facilitator triggers a reveal — no manual counting, no whiteboard tallying needed.
Start Your Next Session
Planning poker is effective precisely because it is simple. The rules are easy to learn, sessions run fast when facilitated well, and the simultaneous reveal consistently surfaces the information a team needs to make good decisions. Whether you are onboarding a new team to agile estimation or looking to improve the quality of an existing process, a well-run planning poker session is one of the highest-return investments you can make in sprint reliability.
Ready to run a session? Corgi Planning Poker is free, requires no account, and your room is ready in seconds.
Start a free planning poker room