Planning poker cards and decks explained

A planning poker deck is the set of values a team is allowed to vote with. Most decks follow a Fibonacci-like sequence such as 0, 1, 2, 3, 5, 8, 13, 20, 40 and 100, plus a few special cards: ? for “not sure”, a coffee cup for “I need a break” and sometimes ∞ for “too big to estimate”. Other popular decks use T-shirt sizes, powers of two or hours.

The values look arbitrary, but the gaps between them are the point. They grow as the numbers grow, because the bigger an item is, the less precisely anyone can size it. This guide explains the common decks, what each special card means and how to choose one for your team. If you are new to the game itself, start with what is planning poker?

Why planning poker cards have gaps

A deck with every number from 1 to 20 invites arguments nobody can win. Is this a 7 or an 8? Nobody knows, and the debate costs more than the difference is worth. A deck with growing gaps forces a more useful question: is this closer to a 5 or to an 8?

Mike Cohn explains the shape with Weber’s law: we notice differences between things as a percentage, not as an absolute amount. You can easily tell 1 kg from 2 kg, but not 20 kg from 21 kg. After 2, each Fibonacci number is about 60% larger than the one before, so every step on the deck is roughly as easy to tell apart as the last. The guide to Fibonacci story points goes deeper into the reasoning.

The gaps were there from the start. James Grenning’s original 2002 deck had only 1, 2, 3, 5, 7, 10 and infinity. His advice to anyone who really wanted to play a 4, 6, 8 or 9 was to put down two cards at once, adding that he doubted the extra precision would help much.

The common planning poker decks

These six decks are the ones you will meet most often. Here is what each looks like and where it fits.

Fibonacci

• 0
• 1
• 2
• 3
• 5
• 8
• 13
• 21
• 34
• 55
• 89

Each value is the sum of the two before it (decks drop the repeated 1). It is the classic deck for story points. Its top values work mainly as warnings: if a 34 or a 55 comes up, the item is usually an epic that needs splitting, not estimating.

Modified Fibonacci

• 0
• ½
• 1
• 2
• 3
• 5
• 8
• 13
• 20
• 40
• 100

This is the deck many people picture when they hear “planning poker”. It keeps the Fibonacci shape for small items and rounds the large ones. Cohn has written that his teams moved from 21 to 20 because 21 implied a precision they could not support, then added 40 and 100 for very large items. The ½ card is for tiny changes that still need doing and testing properly.

Powers of two

• 1
• 2
• 4
• 8
• 16
• 32
• 64

Each card doubles the one before, so the deck is easy to explain (“is this twice as big?”) and rounds are quick. The trade-off is fewer choices in the middle of the range: there is nothing between 4 and 8, or between 8 and 16. Teams whose items vary a lot in size may find the jumps too coarse.

T-shirt sizes

• XS
• S
• M
• L
• XL
• XXL

Sizes instead of numbers. They feel less precise, which is honest for early, rough sizing of a roadmap or a new backlog, and they remove any temptation to read points as hours. The catch is that you cannot add sizes up to plan a sprint or track velocity, so teams that need numbers map each size to a value, for example XS = 1, S = 2, M = 3, L = 5, XL = 8 and XXL = 13. T-shirt sizing covers when to switch to points.

Hours

• 1
• 2
• 4
• 8
• 16
• 24
• 32
• 40

Time-based decks estimate duration directly. They are useful for breaking sprint items into tasks, for operations or support work that comes in predictable chunks, and for client work billed by time. The risks are that each person estimates their own speed, and that an hours estimate is quickly read as a promise. If you use hours, keep them for tasks within a sprint rather than for sizing the backlog.

Linear 1 to 10

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10

A plain linear scale is easy to understand and quick to learn. It also brings back the 6 versus 7 debates the other decks avoid, so it suits teams estimating small, similar items better than teams with a varied backlog.

What the special cards mean

Most decks include a few cards with a special meaning. Some are not sizes at all; others are sizes that carry a particular message.

• ?
• ∞
• 0
• ½

• ? (not sure). “I can’t estimate this with what I know.” It is an honest vote, not a refusal. One ? is a prompt to ask what the person needs to know; several mean the item is not ready to estimate.
• Coffee cup. “I need a break.” If it appears, the session has probably run too long, so take five minutes and come back to the item fresh.
• ∞ (infinity). “This is too big to estimate.” In Grenning’s original deck it meant the customer had to split the story. Treat it as a request to split, not as a number.
• 0. “No meaningful effort”, usually because the work is already done or is covered by another item. Use it sparingly: most changes still need testing and a release.
• ½. A very small change that still has to be done properly, such as a wording fix that needs a release.
• Abstain or pass. “I won’t vote on this one”, for example because the item is in an area you will not touch. Unlike ?, it is a choice to sit out the round rather than a vote.

Agree as a team what the special cards mean and how they count. In particular, decide whether a ? holds up the round or simply prompts a question, and whether people who abstain still join the discussion.

How to choose a deck for your team

Start with what you use estimates for, then pick the simplest deck that serves it.

Three rules apply whichever deck you choose:

• Keep it for months. Changing the deck changes the meaning of every estimate, so past velocity stops being comparable. Change it deliberately, at the start of a sprint, and re-estimate your reference stories when you do.
• Set a ceiling. Agree the largest value that may enter a sprint, such as 13, and treat anything above it as “split me”.
• Use one deck per team. Mixing scales inside a team makes discussion impossible, because a 5 has to mean the same thing to everyone voting.

Physical cards or online cards?

Physical cards are cheap, tactile and need no set-up, which makes them pleasant for a team in one room. Their weaknesses show as soon as the team grows or spreads out: people glance at their neighbours’ cards, someone flips early, the facilitator tallies votes by hand, and remote colleagues have to hold cards up to a camera.

Online cards fix those problems. Every vote stays hidden until the reveal, the tool works out the spread for you, and the estimate can go straight into your backlog. The cost is one more browser tab and a tool to choose: how to choose a planning poker tool has a checklist. For hybrid teams the rule is simple: if anyone is remote, everyone votes online, including the people in the room.

On planningpoker.org The SprintPoker room offers Fibonacci (0 to 89 or 1 to 89), Small Fibonacci (0.5 to 13), Linear (1 to 10), Hours (1 to 40), T-shirt sizes (XS to XXL), T-shirt sizes mapped to numbers, and, in sessions on a Laya board, that board’s own scale. Every deck has a ? card that counts as a vote but never towards the average, and Break and Abstain cards are optional.

Making your own deck

A perfectly good deck can be made from index cards and a marker: one card per value for each estimator, plus ? and any special cards you use. A few practical points:

• Keep it to about 8 to 12 values. More cards mean more debate, not more accuracy.
• Make the gaps grow. Whatever numbers you choose, each step should be a bigger jump than the last.
• Write your reference stories on a card of their own and keep it in the middle of the table.
• Decide what the top card means before anyone plays it: an estimate, or a signal to split.

If you estimate online and want a custom scale, check that the tool supports it before you commit. And if you are unsure where to start, modified Fibonacci is a safe default that most people joining your team will already know.