0, 1, 2, 3, 5, 8, 13… The sequence that conquered both mathematics and sprint retrospectives.
What is the Fibonacci scale in agile?
The Fibonacci scale in agile is a set of estimation values based on the Fibonacci sequence, in which each number is the sum of the two before it. Agile teams use a modified version, usually 1, 2, 3, 5, 8, 13 and 21, often completed by 0, ½ and a "?" card, to size user stories in story points. The scale works because the gaps between values grow with the size of the work: beyond the first few values, each number is roughly 60% larger than the previous one. This matches how estimation behaves in practice, since uncertainty grows with size and people perceive differences in relative rather than absolute terms. Having fewer, clearly distinct values also ends debates about whether a story is a 6 or a 7. Mike Cohn popularized Fibonacci-like decks through Planning Poker. A story estimated at 21 or more is usually too big and should be split.
Fibonacci is one scale among several: our guide to agile estimation techniques for sprint planning compares them all.
Where does the Fibonacci sequence come from?
The sequence is named after Leonardo of Pisa, known as Fibonacci, who described it in his book Liber Abaci in 1202 through a puzzle about breeding rabbits. Each number is the sum of the two previous ones: 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, and so on. As the numbers grow, the ratio between two consecutive values gets closer to the golden ratio, about 1.618 (Wikipedia).
Agile teams drop the duplicate 1 and stop well before the numbers become absurd. Nobody needs a 144-point story.
How did Fibonacci end up in agile estimation?
When James Grenning described Planning Poker in 2002, his original deck was not pure Fibonacci. It held cards for 1, 2, 3, 5, 7 and 10 ideal days, plus infinity. His reasoning already contained the core idea: "As the estimates get longer, the precision goes down." Any story longer than two weeks got the infinity card, and the customer had to split it.
Later decks moved closer to Fibonacci. The deck published by Mountain Goat Software, Mike Cohn's company, uses 0, 1, 2, 3, 5, 8, 13, 20, 40 and 100. Cohn explains that his teams replaced 21 with 20 because 21 implied a precision they could not support, then added 40 and 100 for very large items.
| Scale | Values | Typical use |
|---|---|---|
| Grenning's original deck (2002) | 1, 2, 3, 5, 7, 10, infinity | Release planning in ideal days |
| Pure Fibonacci | 1, 2, 3, 5, 8, 13, 21, 34 | Story points |
| Mountain Goat deck | 0, 1, 2, 3, 5, 8, 13, 20, 40, 100 | Story points, epics |
| Planning Poker Online | 0.5, 1, 2, 3, 5, 8, 13, 21, ?, coffee | Sprint estimation |
| Linear | 1, 2, 3, 4, 5, 6… | Rarely recommended |
| Powers of 2 | 1, 2, 4, 8, 16, 32, 64 | Teams wanting starker gaps |
Why does Fibonacci work for estimation?
1. It matches how we perceive differences. Cohn links the scale to Weber's law: the difference we can detect between two objects is proportional to their size (Weber-Fechner law). Going from 1 to 2 kilograms is a 100% increase that anyone notices. Going from 20 to 21 kilograms is a 5% increase that almost nobody does. Fibonacci values keep a roughly constant relative gap, so each step stays noticeable.
2. Uncertainty grows with size. A small story is well understood, so fine distinctions make sense. A large story hides unknowns, so a coarse scale is more honest. That is also why story points are relative rather than absolute.
3. Fewer choices, faster decisions. With 1 to 10, voters hesitate between 6 and 7. With Fibonacci, the question becomes "is it a 5 or an 8?", which is easier to answer and quicker to agree on.
4. It has a built-in alarm. When cards reach 13 or 21, the scale itself tells the team that the story is too big to be estimated reliably.
Why not use a linear scale like 1, 2, 3, 4, 5?
A linear scale suggests you can tell a 7 from an 8, a 12.5% difference. On complex work, that precision is an illusion, and teams end up spending ten minutes arguing over a distinction that the next sprint will not confirm. Grenning was already making this point in 2002: if you feel the need to play a 4, 6, 8 or 9, combine two cards, but the extra precision will probably not help much.
Powers of 2 are a reasonable alternative for teams that want even sharper choices. Linear scales work only for very small, repetitive items.
Does the Fibonacci scale prevent anchoring?
Not on its own. Anchoring, documented by Tversky and Kahneman (1974), is the tendency to stay close to the first number we hear. In estimation, that first number is usually a colleague's estimate spoken aloud. The scale does not change this, but Planning Poker does: everyone picks a card privately and reveals it at the same time, so nobody hears a number before choosing their own.
In practice, a ticket estimated at 21 or more is a warning signal. Split it before it enters a sprint.
What do 0, ½, ? and infinity mean?
- 0: the story is already done or trivial. No need to discuss it.
- ½: a tiny change, smaller than the team's 1-point reference.
- ? (question mark): the voter lacks information to estimate. Clarify before voting again.
- Infinity: the story is too big to estimate. Split it.
- Coffee cup: the team needs a break, which is a perfectly valid answer after 45 minutes of estimation.
Which mistakes should you avoid with the Fibonacci scale?
- Reading an 8 as 8 hours or 8 days. Fibonacci values are relative story points, not durations.
- Averaging votes. Three votes of 3, 5 and 13 do not make a 7. Discuss the gap and re-vote on the scale.
- Accepting 21s into a sprint. A 21 is a request to split, not a commitment.
- Changing the scale every sprint. Velocity is only meaningful if the values keep the same meaning over time.
Frequently asked questions
Why do some Planning Poker decks use 20 instead of 21? Mike Cohn's Mountain Goat Software deck uses 20, 40 and 100 after 13, rather than continuing the sequence with 21 and 34. Cohn explains that an estimate of exactly 21 implied a precision his teams could not support on large items. Rounder numbers make it clear that big estimates are rough approximations, not exact measurements.
Is the Fibonacci scale mandatory in Scrum? No. The Scrum Guide does not prescribe any estimation scale, nor story points. It only says that the Developers doing the work are responsible for sizing it. The Fibonacci scale is a widespread convention because it reflects growing uncertainty, but teams are free to use T-shirt sizes or powers of 2.
What does an 8 mean on the Fibonacci scale? On the Fibonacci scale, an 8 means a story roughly 60% bigger than a 5 and noticeably smaller than a 13. It is a relative size, not 8 hours or 8 days. For most teams, an 8 is a large story that still fits in a sprint but deserves a closer look at its risks.
Can you convert Fibonacci points to hours? Not reliably. Fibonacci points measure relative effort, and the number of hours behind one point varies from one developer to another. Mike Cohn advises against any fixed conversion. To forecast duration, divide the remaining points by the team's average velocity instead of converting each story into hours.
Sources
- James Grenning, Planning Poker or How to avoid analysis paralysis while release planning, 2002
- Mike Cohn, Why the Fibonacci Sequence Works Well for Estimating, Mountain Goat Software
- Mike Cohn, Planning Poker, Mountain Goat Software
- Amos Tversky and Daniel Kahneman, Judgment under Uncertainty: Heuristics and Biases, Science, 1974
- Ken Schwaber and Jeff Sutherland, The Scrum Guide, 2020
- Fibonacci sequence, Wikipedia
Try the Fibonacci scale with your team
Not convinced by numbers? See how clothing sizes compare in Fibonacci vs T-Shirt Sizing. Ready to try? Create a free Planning Poker session: the Fibonacci cards are ready, just share the link. Discover all the features, including T-shirt and powers-of-2 scales.
