Two scales, two different conversations
Every planning poker tool offers at least Fibonacci numbers and T-shirt sizes. The choice between them is not just aesthetic—it shapes how your team talks about work. Fibonacci forces the team to argue about whether something is a 5 or an 8. T-shirt sizing asks only whether it is medium or large. Both produce estimates; the question is which kind of conversation your team actually needs.
The Fibonacci sequence and what it signals
The planning poker version of Fibonacci—0, 1, 2, 3, 5, 8, 13, 21, 34—is not a precise measurement system. The growing gap between values is intentional: the difference between 8 and 13 is much larger than the difference between 2 and 3. That gap is a signal: the bigger the work, the harder it is to estimate, so the scale spreads out to reflect that uncertainty.
In practice, a "13" should make a facilitator uncomfortable—not because 13 is a bad number, but because large stories carry large uncertainty and are excellent candidates for splitting before the sprint. Teams that estimate many 13s and 21s usually have a story-readiness problem upstream.
What 0 and 0.5 mean
Some Fibonacci variants include 0 (essentially "nothing to do here—it is already done or it is a no-op") and 0.5 (trivial: a one-line fix or a configuration change). These are useful for stories that are technically in the backlog but do not represent real work. Avoid overusing 0—if zero-point stories pile up in a sprint, they are consuming capacity but not appearing in velocity math.
When Fibonacci is the better choice
- You use velocity for sprint capacity planning. If you plan sprints by comparing story points to a rolling velocity average, you need numeric estimates with some precision. T-shirt sizes require an extra mapping step that many teams never actually do.
- You feed estimates into Jira or a similar tracker as story points. Jira's story point field expects a number. Fibonacci maps directly; T-shirt sizes need a translation layer.
- The team uses the debate productively. "Is this a 5 or an 8?" often uncovers a missing edge case or hidden complexity—that disagreement is the value, not the number.
- You need to track estimation accuracy over time. Numeric estimates enable trend analysis: are 5-point stories consistently taking two days? Are 13s frequently overrunning?
T-shirt sizing: speed over precision
T-shirt sizes—XS, S, M, L, XL (sometimes XXL for epics)—trade numeric calibration for coarse, fast bucketing. They answer the question "how big is this roughly?" without implying that the difference between an M and an L is knowable in advance.
The main advantage is accessibility. Non-technical stakeholders who do not live in story points understand "this is a small change" versus "this is a large project" intuitively. T-shirt sizing reduces the anxiety that comes with assigning specific numbers to uncertain work—a new team member who has never done agile estimation can contribute meaningfully.
When T-shirt sizing is the better choice
- Discovery or epic-level grooming. When you are prioritizing themes or features months out, you do not have enough information for Fibonacci precision. T-shirt sizing gives you a rough ordering without false confidence.
- Mixed-audience estimation. If design, marketing, or business stakeholders co-estimate alongside engineers, T-shirt sizes reduce the alienation that numeric story points can create.
- New teams getting started. One less concept to explain. The team can always migrate to Fibonacci once they are comfortable with the ritual.
- When Fibonacci debates become unproductive. If your team spends eight minutes arguing over 5 versus 8 on every story without surfacing genuine insight, T-shirt sizing might cut the noise.
Hybrid patterns that work in practice
Most mature teams land on a hybrid:
- T-shirt for epics and features, Fibonacci for user stories. High-level roadmap items get T-shirts; sprint-level backlog items get Fibonacci. The coarser scale matches the coarser information.
- T-shirts for the first pass, Fibonacci for sprint commitment. Run a quick T-shirt sizing session to prioritize a large candidate backlog; then run a proper Fibonacci session only on the stories entering the next two sprints.
- Parallel scales. Some teams estimate developer effort in Fibonacci and design effort in T-shirts, keeping both visible in their tracker.
Whatever pattern you adopt, document it and stick to it for at least a quarter before evaluating. Teams that constantly switch scales cannot calibrate, cannot track velocity meaningfully, and lose trust in their own estimates.
The modified Fibonacci variants worth knowing
Beyond the classic sequence, a few variants come up in practice:
- Fibonacci Flat (0, 0.5, 1, 2, 3, 5, 8, 13, 20, 40, 100): Includes the "?" and "∞" wildcards in many tools. The 20 and 40 replace 21 and 34 for easier mental math; 100 signals "this is not a story, this is a project."
- Doubled sequence (1, 2, 4, 8, 16, 32, 64): Clean powers of two. Common in teams with a software engineering background that prefers binary thinking.
- Hours (2, 4, 8, 16, 24): When the team genuinely wants to estimate in hours—usually for short-cycle work like ops tickets or design tasks where effort is more predictable.
The "?" and coffee / banana cards
Beyond estimation cards, most poker tools include special cards:
- "?" — "I don't have enough information to estimate this story." Forces a clarification conversation. Useful; use it rather than guessing.
- "∞" — "This story is too large to estimate meaningfully." Signals that the story needs to be split before it goes anywhere near a sprint.
- Coffee / banana — "I need a break." Playful cards that let team members signal fatigue without disrupting the meeting. After 90 minutes of estimation, these show up more.
How to introduce a new scale to your team
Switching estimation scales mid-project causes confusion. If you want to migrate from T-shirts to Fibonacci (or vice versa):
- Announce the change before the next refinement; explain why—velocity tracking, Jira integration, or stakeholder clarity.
- Establish reference stories under the new scale using 2–3 stories the team has recently completed.
- Run the first session with extra patience for calibration questions.
- Do not re-estimate old stories in the new scale. Accept the historical break and move forward.
Try both decks in ScrumJam
ScrumJam ships Fibonacci, Fibonacci Flat, T-shirt, Hours, Doubled, and Simple Numbers decks on the free path. Custom decks with your own values and labels are available on the lifetime tier. Open the planning poker picker, choose your deck, and run a dry session with three or four real stories. Pair this with the sprint planning poker guide for facilitation detail and the agile scrum estimation practices article for deeper technique.