Choosing an estimation scale: Fibonacci, T-shirts, or five fingers?
BESTIMATE ships with eleven card decks, and that number raises a fair question: why would anyone need more than one way to say "this is medium-sized"? Because the scale you estimate on quietly decides what conversation your team has.
Growing gaps mean honest uncertainty
Fibonacci (1, 2, 3, 5, 8, 13, 21 …) dominates for a reason: the gaps grow with the numbers. Whether a story is a 2 or a 3 is worth discussing; whether it is a 20 or a 21 is not, and the scale physically prevents that debate. Values from 13 upward stop being estimates and start being diagnoses: the story is too big — split it. A sequential 1–10 scale feels friendlier to newcomers but invites exactly the sham precision Fibonacci avoids; powers of two go the other way and force one brutal question per card: "twice as big as that?"
Sometimes numbers are the problem
T-shirt sizes (XS–XXL) exist because nobody can do maths with letters — no averages, no false precision, no velocity charts built on sand. That makes them ideal early: roadmaps, epics, stakeholder conversations. Ideal days sit at the opposite end — an absolute scale for time-and-budget planning that stakeholders understand instantly, and precisely because of that, one to keep away from sprint velocity.
Fist of Five isn't an estimation scale at all — it's a consensus thermometer. Zero to five fingers, shown simultaneously, answer "how confident are we?" better than any number. Teams use it as a confidence vote on sprint goals, architecture decisions, release readiness.
Scales for volume, scales for money
A hundred unestimated backlog items need a different tool than nine refinement stories. The bucket scale (0–200) exists for exactly that volume — sort relative to a reference story, fast. Magic Estimation does it in silence. And when the estimate feeds a quote or a budget, three-point / PERT replaces one number with three (optimistic, likely, pessimistic) and a formula that admits uncertainty instead of hiding it: E = (O + 4M + P) / 6.
Finally, NoEstimates refuses the question: slice everything to a similar size, count instead of estimating, forecast from throughput. Its deck has three cards — ✓ small enough, ✂ split it, ? unclear — because that is genuinely the whole conversation.
The full guide, with a comparison table and a chapter per method, lives at bestimate.buddycore.com/guide.