What is an Agile estimation scale?
An estimation scale is the shared set of labels a team uses to compare backlog items. The labels may be numbers or categories; they are relative reference points, not a conversion table to hours. Scrum itself does not prescribe Story Points, Planning Poker or a particular scale, so the team should choose and calibrate its own approach.
Agile estimation scales compared
| Scale | Values | Best for | Main characteristic |
|---|---|---|---|
| Fibonacci | 0, 1, 2, 3, 5, 8, 13, 21 | Story-level estimation | Gaps grow with uncertainty |
| Modified Fibonacci | 0, 0.5, 1, 2, 3, 5, 8, 13, 20, 40, 100 | Mixed-size backlogs | Adds broad high-end buckets |
| T-Shirt Sizes | XS, S, M, L, XL, XXL | Early, coarse sizing | Fast categorical comparison |
| Powers of 2 | 1, 2, 4, 8, 16, 32, 64 | Teams wanting clear separation | Every step doubles |
| Linear 1–10 | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 | New teams and simple workshops | Familiar but more granular |
Fibonacci
The familiar 0, 1, 2, 3, 5, 8, 13 and 21 deck gives small work fine distinctions and larger work wider gaps. Those gaps acknowledge that uncertainty usually grows with size. Treat 13 or 21 as a prompt to ask whether the item should be split, not as a promise of duration.
Modified Fibonacci
This deck adds 0.5 and replaces the distant high end with 20, 40 and 100. It is useful when a team sizes tiny changes and coarse initiatives in one workshop. A 40 or 100 should normally trigger decomposition before sprint planning.
T-Shirt Sizes
XS through XXL works well when speed matters more than arithmetic: roadmap shaping, discovery or an initial pass over epics. The values are ordered, so mode, distribution and spread are meaningful; a numeric average is not.
Powers of 2
A 1, 2, 4, 8, 16, 32 and 64 deck forces a clear choice: each card is twice its neighbour. It suits teams that prefer strong separation, although discussion can drift into whether work is literally twice as large.
Linear 1–10
A linear scale is easy to explain and can suit workshops where participants are new to relative estimation. Because adjacent choices are close, teams should avoid debating 6 versus 7 as if either were precise.
Fibonacci vs Modified Fibonacci
Both widen gaps as size grows. Standard Fibonacci is compact and encourages earlier splitting; Modified Fibonacci preserves very small and very large buckets for broader backlog work.
Fibonacci vs T-Shirt Sizes
Fibonacci supports numerical Story Points and velocity trends. T-Shirt Sizes are intentionally coarser and often faster. Teams can size epics with T-shirts, split them, then estimate resulting stories with Fibonacci.
Fibonacci vs Powers of 2
Both reduce false precision. Powers of 2 creates stronger separation, while Fibonacci offers more choices near the small end. Neither is inherently more accurate; consistency and reference items matter more.
Which estimation scale should your team use?
Start with Fibonacci when you want a familiar default. Use Modified Fibonacci when very large items must remain visible before splitting. Choose T-Shirt Sizes for fast, coarse portfolio or discovery work. Powers of 2 make size differences unmistakable. Linear 1–10 feels intuitive, but adjacent values can invite false precision. Keep one scale long enough to build stable reference stories, and revisit it only when it stops helping the conversation.
How to use these scales in PokerScrum
Choose the estimation method before creating the room. Every participant receives the same deck. Votes stay private until reveal; then discuss the highest and lowest estimates, clarify assumptions and vote again if new information changes anyone’s view. The question-mark and coffee cards remain available for uncertainty and pauses.
Create a free Scrum Poker room