What is planning poker?
A short guide to the technique itself — the round, the decks, confidence levels, why relative sizes beat hours, and the ways it goes wrong.
Planning poker is a consensus estimation technique. The team sizes a piece of work relative to other work rather than in hours: everyone picks a card privately, all cards turn over at once, and the differences are the point. A number that surprises the room usually means somebody knows something the others do not.
People are poor at absolute duration and good at comparison. “About twice the last one” is a judgement a team can make honestly, and over a few sprints those points turn into a velocity that predicts better than any hour estimate given under pressure.
One person talks first and everyone follows. The story is really five stories. The number becomes a promise instead of an estimate. Somebody treats a point as a day. The technique only works while the vote stays private and the discussion stays short — which is exactly what this tool enforces.
Fibonacci 1 2 3 5 8 13 21 — gaps widen as certainty drops.
Full Scrum 0 ½ 1 2 3 5 8 13 20 40 100.
T-shirt XS to XXL, for teams that would rather not see numbers.
Custom — your own values, in your own order: hours, days, XS–XXL with a “spike” card, or a scale that matches your team’s history. Set once per room and saved to the organisation for reuse (Pro).
? and ☕ — “I need more information” and “I need a break”. Both are valid answers, on every deck.
Described by James Grenning in 2002 and popularised by Mike Cohn, planning poker is a variant of the Wideband Delphi method — expert estimates gathered independently, then reconciled through discussion.
Ten to fifteen stories in an hour is healthy. If a single story eats ten minutes, stop estimating it and write down the question that is blocking you.
A number on its own hides how much the team actually knows. After a wide spread, everyone answers one extra question — how sure are you about this estimate — and the answer is stored with the story.