Fibonacci in planning poker: Why teams use it

Planning poker teams vote with the Fibonacci sequence (0, 1, 2, 3, 5, 8, 13, 21) because the gaps between the numbers grow as the numbers get bigger, which matches a simple truth about estimating: the larger a piece of work is, the less precisely anyone can size it. A growing scale stops the team from agonising over whether something is a 13 or a 14, and forces a clear choice instead. Below: what the sequence is, where the practice came from, the real reason it works, and when it gets in the way.

What is the Fibonacci sequence?

The Fibonacci sequence is a list of numbers where each one is the sum of the two before it: 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, and so on. It shows up all over mathematics and nature. The detail that matters for estimation is the spacing: as you go further along, each number is roughly 1.6 times the one before it (a ratio that approaches the golden ratio). So the jumps start small and get steadily larger.

The Fibonacci scale in planning poker

Teams do not use the raw sequence. A typical planning poker deck drops the repeated 1 and looks like this:

0, 1, 2, 3, 5, 8, 13, 21, plus a ? (I do not know) and a coffee cup (I need a break).

You pick the card closest to your sense of the item's size, everyone reveals at once, and you discuss the spread. If you are new to the ritual, see our step-by-step guide to running planning poker.

Where it came from

Planning poker was first defined and named by James Grenning in 2002, in a short paper written while he was looking for a lighter alternative to Wideband Delphi, an older group estimation method. Mike Cohn later popularised the technique in his book Agile Estimating and Planning, and the Fibonacci scale travelled with it. So the practice is about two decades old, and the Fibonacci choice has been the common default for most of that time. (References in Sources below.)

Why the gaps grow (the real reason)

The point of a widening scale is to stop pretending you have precision you do not have.

Think about estimating a small, well-understood task. The team can reasonably tell a 2 from a 3. Now think about a large, fuzzy one. Can anyone honestly tell a 21 from a 22? No. The uncertainty in a big item is far larger than one unit. A linear scale (1, 2, 3, 4, 5, 6, 7, 8) invites endless debate over neighbouring numbers that the team cannot actually distinguish. The Fibonacci gaps remove those false choices: past 13 your only options are 21, then 34, so you commit to an order of magnitude instead of haggling over a point.

This is also why many teams round the big numbers (more on that below). A 40 does not claim to be exact. It says "big, and we are not sure how big".

The deeper reason: how people judge size

There is a well-studied reason a growing scale fits human estimation. In psychophysics, Fechner's law holds that our perception of a stimulus grows with the logarithm of its actual intensity: we sense the world in ratios, not absolute steps. A light that is twice as bright, or a sound twice as loud, feels like one step up, whether you started dim or bright. Experiments suggest the same logarithmic pattern applies when people judge quantities and numbers, and one analysis argues a logarithmic scale is optimal because it keeps the relative error roughly constant across the range.

A Fibonacci-style scale, where each step is about 1.6 times the last, is close to that natural logarithmic spacing. In plain terms: the scale lines up with how our minds already compare sizes, so the steps feel evenly spaced even though the raw numbers are not. That is the honest version of "Fibonacci just feels right".

Modified Fibonacci and other scales

The pure sequence gets awkward at the top (34, 55, 89 are hard to hold in your head), so a widely used variant rounds them off:

Modified Fibonacci: 0, ½, 1, 2, 3, 5, 8, 13, 20, 40, 100.

The rounder large numbers make the same point as the gaps do: at that size you are naming a ballpark, not a figure. Beyond Fibonacci, teams also use T-shirt sizes (XS to XXL) for quick rough sizing and powers of two (1, 2, 4, 8, 16) for an even steeper curve. The exact scale matters less than two things: that the choices are limited, and that the team uses the same one consistently so its numbers stay comparable over time.

When Fibonacci helps, and when it gets in the way

It helps when:

• The team keeps slipping into hour-by-hour debate. The gaps force a decision and keep sessions moving.
• Items vary a lot in size. A scale that spans 1 to 100 handles both tiny and large work.

It gets in the way when:

• Everything lands on 13 or 21. That is not a scale problem, it is a signal the items are too big. Split them before estimating.
• The team treats the number as exact. Fibonacci is a convenience for making a rough call, not a measurement. If people start converting a 5 into "ten hours", the scale is being misused (see what story points are, and are not).
• Work is uniform and well understood. If every item is genuinely similar, you barely need a scale at all.

Being clear-eyed about this matters: the sequence is a useful default, not a source of accuracy on its own.

How to use the Fibonacci scale well

• Anchor with a reference story. Agree that one familiar item is your 3, and compare everything to it. This keeps the scale stable across sessions.
• Keep the same deck. Switching scales mid-project makes your history meaningless.
• Use ? and the coffee card. "I do not know" is real information. It usually means the item needs clarifying before it can be sized.
• Re-estimate when you learn more. A number chosen before the work was understood is not binding.

Fibonacci vs. other scales

Frequently asked questions

Why do planning poker cards use the Fibonacci sequence?

Because the gaps between the numbers grow, which matches the fact that bigger items are harder to size precisely. The widening steps stop the team from debating numbers it cannot actually tell apart, like 13 versus 14.

What are the Fibonacci numbers used in planning poker?

A common deck is 0, 1, 2, 3, 5, 8, 13, 21, plus a "?" card for unknown and a coffee card for a break. A modified version rounds the large numbers to 20, 40, 100.

Is Fibonacci better than a 1 to 10 scale?

For most teams, yes, because a linear 1 to 10 scale encourages arguments over neighbouring numbers that nobody can genuinely distinguish. Fibonacci's gaps force a clearer, faster choice.

What is modified Fibonacci?

A rounded variant, 0, ½, 1, 2, 3, 5, 8, 13, 20, 40, 100. It swaps the awkward large Fibonacci numbers (34, 55, 89) for rounder ones, which also signals that large estimates are approximate.

Does the science actually support Fibonacci?

Indirectly. Research on perception shows people judge magnitude on a roughly logarithmic, ratio-based scale, and a Fibonacci-style scale (each step about 1.6 times the last) is close to that natural spacing. It fits how we compare sizes, though no single scale makes estimates accurate by itself.

What do the ? and coffee cup cards mean?

The "?" means you cannot estimate the item yet, usually because it needs clarifying. The coffee cup is a light way to ask for a break.

Sources

1. Grenning, J. (2002). Planning Poker or How to Avoid Analysis Paralysis While Release Planning. (the paper that first defined and named the technique).
2. Cohn, M. (2005). Agile Estimating and Planning. Prentice Hall (popularised planning poker and the Fibonacci scale).
3. Fechner, G. (psychophysics; Weber-Fechner law): perception of intensity grows with the logarithm of the stimulus. See overview and the logarithmic account of number perception.
4. Portugal, R. & Svaiter, B. (2011). Weber-Fechner Law and the Optimality of the Logarithmic Scale. Minds and Machines (a logarithmic scale minimises the maximal relative error).