Fibonacci, T-shirt sizes or 1 to 10: picking a planning poker scale

Every planning poker session starts with a choice most teams make once and never revisit: which cards are in the deck. The scale looks like a detail, but it shapes how people think about size, how much they argue about small differences, and how easy it is to say "I have no idea".

This post compares the three most common scales, Fibonacci story points, T-shirt sizes and a linear 1 to 10, and explains when each one fits.

What a scale is for

An estimation scale is not a clock. Story points do not convert to hours, and they are not meant to. They express relative size: this story is about as big as that one, or roughly twice as big. A good scale makes those comparisons easy and makes false precision hard.

Two properties matter most:

• Gaps that grow with size. The bigger a piece of work, the less precisely anyone can size it. A scale should not ask people to choose between 14 and 15.
• A way to say "unknown". If the only options are numbers, people who cannot estimate will still pick one, and that number will look just as confident as the others.

Fibonacci story points

The classic planning poker deck uses a Fibonacci-like sequence: 0, 1, 2, 3, 5, 8, 13. Each step is roughly the sum of the two before it, so the gaps widen as the numbers grow.

Why teams like it:

• The widening gaps match how uncertainty grows with size. Nobody debates 8 against 9, because 9 is not a card.
• It forces a decision at the top. A story that does not feel like a 13 is a signal to split it, not to invent a 21.
• Velocity and history work well with numbers, so it is easy to compare sprints and spot stories the team keeps getting wrong.

Watch out for: treating points as hours, or adding a 20, 40 and 100 to the deck so that big stories never get split. A short scale is a feature.

T-shirt sizes

T-shirt sizing replaces numbers with XS, S, M, L and XL.

Why teams like it:

• It is fast and intuitive for people who are new to estimation, including stakeholders and guests.
• It discourages arithmetic. Nobody tries to add an M to an L, which keeps the focus on relative size.
• It works well for early roadmap sizing, where the question is "small, medium or huge?" rather than "3 or 5?".

Watch out for: velocity. Sizes do not sum, so if you need sprint capacity planning you will end up mapping sizes to numbers anyway. Many teams use T-shirt sizes for epics and Fibonacci for stories.

Linear 1 to 10

A linear scale offers every number from 1 to 10.

Why teams like it: it is familiar, it needs no explanation, and it can suit teams that score something other than effort, like complexity, risk or confidence on a 1 to 10 scale.

Watch out for: precision arguments. With ten evenly spaced options, people debate 6 against 7 for stories nobody can size that precisely. For effort estimation, a scale with widening gaps usually produces faster, calmer meetings.

The ? and infinity cards

Whatever scale you pick, keep two special cards in the deck:

• ? means "I do not know enough to estimate this". It is the most underused card in planning poker. A room full of question marks is a clear sign the story needs more refinement, not a vote.
• Infinity means "this is too big to estimate as one story". It is a polite way of saying "split it".

Neither card should drag the average around. In EstimateTask both are left out of the average, so someone saying "I do not know" does not quietly turn into a vote of 0 or 100.

How to choose

A simple rule of thumb:

• Sprint stories with a team that tracks velocity: Fibonacci.
• Roadmap sizing, epics or sessions with many non-engineers: T-shirt sizes.
• Scoring something that is not effort, like risk or confidence: linear 1 to 10.

Whichever you choose, keep it for a while. Estimates are only comparable within a scale, so switching every few sprints throws away the history that makes estimation useful.

Scales by role

Scale choice matters even more when the team votes in role groups. With Dev, QA and DevOps averages shown side by side, a scale with widening gaps makes real disagreement stand out: a Dev 3 against a QA 8 is two steps apart on Fibonacci, which is hard to ignore. On a linear scale the same disagreement can hide as "only a few points".

Scales in EstimateTask

Every plan includes three built-in scales: Fibonacci (0, 1, 2, 3, 5, 8, 13), Linear 1 to 10 and T-shirt (XS to XL), each with ? and infinity. The Team and Business plans add custom scales, so you can build your own deck. Scales never change in place, so a past session keeps the values it was voted with.

Read more about why role-specific estimates beat one number, compare pricing, or start free and pick your deck.