What is planning poker?

Planning poker (also called Scrum poker) is a way for an agile team to estimate the effort of a piece of work together. Everyone picks an estimate in private, the whole team reveals at the same moment, and the people furthest apart explain their thinking before the team settles on a number.

How a round works

1. Present the story. The product owner or facilitator reads the user story and answers questions about it.
2. Everyone votes in private. Each estimator chooses a card that represents their estimate, usually in story points.
3. Reveal together. All cards are shown at once, so no one is influenced by the others before deciding.
4. Discuss the outliers. If the votes differ, the highest and lowest voters say why. They often know something the rest of the team doesn't: a hidden dependency, an existing component that can be reused, a risky migration.
5. Re-vote or agree. The team votes again, or agrees on a number, and the facilitator records the final estimate.

Why hide the votes until the reveal?

When estimates are spoken out loud one at a time, the first number said tends to pull every later estimate toward it. This is called anchoring. A senior developer saying "that's a 3" can quietly stop a junior developer from mentioning the reason they thought it was a 13. Revealing every vote at the same moment means each person has committed to their own estimate first, and the disagreement becomes visible instead of being smoothed over.

What do the cards mean?

Most teams estimate in story points: a relative measure of size that combines effort, complexity and uncertainty. A 5 should feel about as big as other 5s the team has done, and noticeably bigger than a 2. Points are not hours; their value comes from comparing stories with one another.

Decks use a sequence with growing gaps, such as the modified Fibonacci sequence (0, 1, 2, 3, 5, 8, 13, 20, 40, 100), because the bigger a piece of work is, the less precisely anyone can estimate it. Some teams prefer T-shirt sizes (XS, S, M, L, XL) for rough, early estimates. See estimation scales compared for when to use each.

Decks usually also include a few special cards:

• ? means "I don't know enough to estimate this yet."
• ∞ means "this is too big to estimate; split it."
• ☕ means "I need a break."

Tips for running planning poker well

• Agree on a reference story. Pick a small, well-understood story the whole team knows and call it a 2 or a 3. Estimate everything else relative to it.
• Timebox the discussion. If two rounds of voting haven't converged, the story probably needs splitting or more information, not more debate.
• Only the people doing the work vote. The product owner explains the story; the developers and testers estimate it.
• Don't average away disagreement. An average of 2 and 13 is not a useful estimate. Talk about why the votes were so far apart.
• Split anything too big. If the team keeps reaching for 40, 100 or ∞, the story is an epic.

Where planning poker came from

The technique was described by James Grenning in 2002 and popularized by Mike Cohn's book Agile Estimating and Planning. It builds on the Wideband Delphi method, in which experts make anonymous estimates, discuss them, and repeat until the estimates converge.