Fibonacci Story Points, Explained

Most planning poker decks don't count up one at a time. Fib's default deck is 0, 1, 2, 3, 5, 8, 13, 21. After the first few cards, each number is the sum of the two before it, so the gaps widen as the numbers grow. Those gaps do most of the work.

Why the gaps grow

Telling a 2 from a 3 is easy. Both are small, and you can picture all of the work. Nobody can honestly tell a 14 from a 15, because at that size the unknowns swamp the difference. A linear scale invites that kind of false precision. The Fibonacci scale removes it by only offering numbers that are far enough apart to mean something.

Past the first few cards, each step is roughly 60 percent bigger than the one before. Choosing between neighbours is always a real decision: is this closer to an 8, or closer to a 13? A team can answer that in a minute.

Reading the big numbers

The bigger the number, the less it says about the work and the more it says about what the team doesn't know yet. A rough guide most teams end up with:

  • 1 to 3: small and well understood. The team could list every step.
  • 5 to 8: a normal piece of sprint work, with a few unknowns the team expects to resolve along the way.
  • 13: large, with real uncertainty. Worth asking whether it can be split.
  • 21: too big to estimate with any confidence. Split it before it goes into a sprint.

So when a team votes 13 and 21, don't spend the time picking a winner. Ask what nobody knows yet. A short spike to find out usually does more for the estimate than another round of voting.

The modified Fibonacci scale

The modified scale, 0, ½, 1, 2, 3, 5, 8, 13, 20, 40, 100, keeps the small numbers, adds a ½ for trivial changes, and swaps the strict sequence at the top for rounder numbers. Those big cards are for sizing epics and roadmap items, where nobody pretends to know whether something is a 34 or a 40. Fib ships both decks, and the host can switch between them from the room settings.

Zero and the special cards

A 0 means the work is already done, or so small it isn't worth tracking. The special cards sit outside the scale: ? when you can't estimate yet, ☕ for a break, and 🃏 when the story is bigger than the biggest card. A 🃏 is a stronger signal than a 21. It says the story has to be broken down before anyone can point it.

Points are relative, not hours

A story point only means something next to other stories the same team has pointed. Your 8 is not another team's 8, and that's fine. What makes points useful is consistency: over a few sprints, the team learns roughly how many points it finishes, and that number becomes the forecast.

Fibonacci estimation questions

Why not estimate in hours?
Hours suggest a precision most estimates don't have, and they change depending on who picks up the work. Points ask a simpler question: is this bigger or smaller than that? Teams answer it faster and more consistently.
What estimation scale does Fib use?
Rooms start on the Fibonacci deck: 0, 1, 2, 3, 5, 8, 13, 21. The host can switch to the modified Fibonacci scale, powers of two, or T-shirt sizes at any time, and every deck carries the ?, ☕ and 🃏 cards.
Is 21 a valid estimate?
It's a valid vote, but it's telling you to split the story. Something that big usually hides several smaller pieces, and splitting it gives better estimates than arguing about whether it's a 21 or a 34.
Can we use other numbers than Fibonacci?
Yes. What matters is that the gaps grow with size. Powers of two work the same way, and T-shirt sizes drop numbers altogether. Pick one scale per team and stick with it, so this sprint's estimates compare with last sprint's.
Support Fibv2026.11.1