Planning Poker Card Decks Explained - Which Scale Should Your Team Use?

Most teams pick a planning poker deck once, on the first day, based on whatever the tool defaulted to — and then never revisit it. That's usually fine. But if your estimates feel imprecise, or your refinement sessions run long, the deck is worth a second look, because different scales are genuinely better at different jobs.

Here's what each common deck is for.

The decks at a glance

Fibonacci — the default for a reason

1, 2, 3, 5, 8, 13

If you don't have a specific reason to use something else, use this. The widening gaps between values reflect the fact that estimation gets less precise as work gets bigger, and the limited number of options keeps the decision fast.

Its main weakness is at the extremes. It's poor at estimating very small work, because everything under a day collapses into "1". And it's poor at epic scale, because the sequence runs out well before you get to quarter-sized initiatives. More on how the Fibonacci scale works.

Choose it if: you're estimating user stories for a sprint. Which is most teams, most of the time.

Default — Fibonacci with room at the bottom

0, 0.5, 1, 2, 3, 5

The same idea, shifted downward. Adding 0 and 0.5 gives you somewhere to put work that's real but tiny — a config change, a copy fix, something already half-done by a previous story.

The 0 card is more useful than it looks. It lets a team acknowledge a task exists and needs doing without inflating the sprint's point total, which keeps velocity meaningful.

The risk is that 0 and 0.5 attract debate wildly out of proportion to their value. Ten minutes arguing whether something is a 0 or a 0.5 is ten minutes you will not get back.

Choose it if: a meaningful share of your backlog is genuinely trivial work.

Half — for when everything is roughly the same size

0.5, 1, 1.5, 2, 2.5, 3

A narrow, evenly-spaced scale. It deliberately breaks the widening-gap principle, which makes it a poor general-purpose choice — but a good one for a specific situation: teams whose work is uniform and well understood, where the real distinctions are small.

Maintenance teams, support rotations, and teams working through a long queue of similar tasks often find Fibonacci too coarse. Half cards give them resolution where they need it.

Choose it if: your stories are consistently small and similar, and Fibonacci keeps flattening real differences.

Large — for epics and quarters

0, 20, 40, 60, 80, 100

Trying to estimate a quarter-sized initiative on a 1–13 scale doesn't work; everything is a 13. A separate large-value deck gives you a scale built for that altitude.

The important discipline is keeping the two scales apart. A 40 here is not "40 story points" in the sense your sprint velocity uses — it's a rough size on a different ruler. Mixing them corrupts velocity, which is the metric most likely to be quoted at you in a stakeholder meeting.

Choose it if: you're doing roadmap or quarterly planning rather than sprint refinement.

T-shirt sizing — for when numbers get in the way

XS, S, M, L, XL, XXL

Numbers invite false precision and, worse, invite people to convert them into days. T-shirt sizes resist both. Nobody asks how many hours an M is.

This makes them excellent early in product discovery, when requirements are vague, and excellent with non-technical stakeholders, who tend to find sizes more intuitive than an arbitrary point scale.

The trade-off is that you can't compute velocity from letters. Many teams use T-shirt sizes during discovery and convert to points once stories are refined — XS becomes 1, S becomes 3, M becomes 5, L becomes 13, XL becomes 20 or a split.

Choose it if: you're sizing early, roughly, or with people outside the delivery team.

Custom decks — use sparingly

Sometimes the standard sets genuinely don't fit. A ? card for "I don't have enough information to estimate this" is the most common and most useful addition — it turns an unproductive guess into an explicit request for clarification, which is almost always the better outcome.

Some teams add a coffee cup as a light-hearted "I need a break", which sounds frivolous but functions as a real pressure valve in a long refinement session.

Beyond that, be cautious. Every extra card is another decision for six people to make on every story. The value of a small deck is speed, and custom decks are where that speed usually goes to die.

So which should you use?

For most teams: Fibonacci, and stop thinking about it.

Switch to Default if a lot of your work is tiny. Switch to Half if your work is uniform and Fibonacci feels blunt. Keep a Large deck aside for quarterly planning. Reach for T-shirt when you're sizing early or explaining to stakeholders.

And if your refinement sessions are painful, change the deck last. It's usually not the deck — it's stories that arrived under-refined, or a facilitator averaging the votes instead of asking why they differed.

All five decks are built into PlanningPoker.live, and you can switch between them mid-session or define your own. Start a free session and try a couple.