All postsEstimation Insights

Fibonacci vs T-Shirt Sizing: Which Estimation Scale Should Your Team Use?

March 8, 20255 min read
📏

Pick the wrong estimation scale and your team will fight the tool instead of using it. The difference between Fibonacci and T-shirt sizing isn't aesthetics — it's about how precise you actually need your estimates to be.

Fibonacci (1, 2, 3, 5, 8, 13, 21)

Fibonacci is the default for a reason. The widening gaps between numbers mirror how uncertainty grows with story size. The jump from 5 to 8 is meaningful; the jump from 21 to 22 is noise — so the scale stops giving you false precision.

Best for: mature teams that already know their velocity, codebases where stories are roughly comparable sprint to sprint, and any team that wants to forecast multi-sprint roadmaps using point totals.

T-shirt sizing (XS, S, M, L, XL)

T-shirt sizes remove the urge to debate whether something is a 5 or an 8. They're qualitative — which is exactly right when you don't yet have enough data to be quantitative.

Best for: new teams without an established velocity, very early backlog discovery (epics, themes), discovery work where requirements are still vague, and cross-team prioritisation conversations where one team's 5 isn't another team's 5.

Powers of two (1, 2, 4, 8, 16)

A less common middle ground. Cleaner math than Fibonacci, but the gap from 8 to 16 is a cliff — teams often end up wanting a number in between and can't have one. Use it only if your team specifically wants to discourage estimating large stories without breaking them down.

How to decide

  • If you can name three stories your team has finished that are roughly the same size as the one you're estimating → use Fibonacci
  • If you can't, or you're estimating an epic → use T-shirt sizing
  • If you find yourselves arguing 5 vs 8 every story → your stories are too big, not your scale wrong
  • Don't mix scales in the same backlog — pick one and stick with it for at least a quarter

The honest answer

Most teams should start with T-shirt sizing for the first 2–3 sprints, then graduate to Fibonacci once they have enough completed stories to anchor against. Trying to use Fibonacci on day one is how teams end up with meaningless point totals that no one trusts.