What is planning poker?

The short version

• Each estimator holds a deck of cards with numbers on them, usually a Fibonacci-like sequence.
• The team discusses one item of work, then everyone picks a card privately.
• All cards are revealed at once, so nobody anchors on the loudest or most senior voice.
• Matching estimates get recorded. Different estimates are not a problem; they are the point. The outliers explain their reasoning, and the team votes again.
• All you need is a deck of cards and your team, or a free online room if you're remote.

Where it comes from

Planning poker was described by James Grenning in 2002 and popularized by Mike Cohn's book Agile Estimating and Planning. It is a lightweight descendant of Wideband Delphi, an estimation method built on a simple observation: a group of practitioners, estimating independently and then discussing, produces better forecasts than any single expert, and far better than a group that estimates out loud, where the first number spoken drags everyone else toward it.

How a session works

1. The facilitator brings the list of items to estimate: user stories, features, tasks.
2. Someone who knows the item (often the product owner) presents it briefly, and the team asks clarifying questions.
3. Every estimator picks a card, face down. Nobody sees anyone else's choice.
4. All cards are revealed simultaneously.
5. If the numbers agree, the estimate is recorded and the team moves on. If they don't, the highest and lowest estimators each explain their reasoning, and the team votes again.
6. Most items converge within a round or two. If one doesn't, that disagreement is useful information (more on that below).

A whole backlog rarely needs estimating in one sitting. Timebox the session, keep discussion per item to a few minutes, and park anything that turns into a design debate.

Why the simultaneous reveal matters

The card mechanic isn't a gimmick. It neutralizes two forces that quietly ruin estimates made out loud:

• Anchoring. Once someone says “that's about a five,” every later estimate gravitates toward five. Private voting means every estimate is formed independently.
• Hierarchy. When the tech lead speaks first, disagreeing has social cost. When cards flip together, the junior developer's 13 gets the same weight as everyone's 5, and it forces the question that matters: what do they know that we don't?

That question is the real output of planning poker. A spread of estimates almost always means people are imagining different work: one person assumed the old API could be reused, another knew it has to be rewritten. The estimate is just the excuse to surface those assumptions before the sprint starts instead of two weeks into it.

Why the cards use Fibonacci numbers

Most decks use a Fibonacci-like sequence (1, 2, 3, 5, 8, 13, 21) rather than every integer. The growing gaps are deliberate: people are decent at telling a 2 from a 3, but nobody can meaningfully distinguish a 12 from a 13. The bigger the work, the bigger the uncertainty, and the sequence forces that honesty. You can't hide behind false precision; you have to choose between 8 and 13.

Common variants include T-shirt sizes (XS to XL) for rough, early-stage sizing and powers of two (1, 2, 4, 8, 16). Most decks also carry a ? card (“I can't estimate this without more information”) and a coffee-cup card (“we need a break”). Both are worth taking seriously when they appear.

Story points, not hours

Planning poker estimates are usually story points: relative measures of size, complexity, and uncertainty, not units of time. A 5-point story is roughly as big as other 5-point stories the team has done; that is all a point claims. Over a few sprints, the team's measured velocity converts points into a calendar forecast far more reliably than summing hour guesses ever does. Hours invite false precision, vary by who picks up the task, and have a way of getting compared against timesheets, at which point people stop estimating and start negotiating.

Common questions

Is planning poker only for Scrum teams?

No. It comes from the Scrum and XP world, but any team that needs to size work before committing to it can use planning poker: Kanban teams, support teams splitting projects, even non-software groups.

How many people should take part?

Everyone who will actually build the work should vote, typically three to ten people. Stakeholders can watch as observers, but estimates belong to the people doing the work.

Does planning poker work for remote teams?

Yes. Online planning poker rooms replace the physical card deck: everyone votes privately from their own device and the cards are revealed simultaneously, exactly like sitting around a table.

What if the team never converges on one number?

After two or three rounds, either take the higher estimate and move on, or treat the disagreement as a signal: the story is probably too big or too vaguely defined, and should be split or clarified before anyone works on it.

Are the estimates commitments?

No. Estimates are forecasts made with the information available at the time. They help a team plan a sprint and spot risk early. They are not deadlines, and treating them as deadlines quickly teaches people to inflate them.

Try it with your team

Point Taken is a free planning poker room built for exactly this: live voting, cards that stay hidden until the reveal, and nothing between your team and an estimate. No sign-up, no ads, no tracking. Open a room, share the link, and start with your next backlog item.