T-shirt sizing in agile: how it works and when to use it

What t-shirt sizing is, how to run a session, and how to turn S/M/L into story points, plus when planning poker is the better call.

Most estimation techniques ask for a number. T-shirt sizing asks a simpler question first: is this thing small, medium, or large compared to the work around it? That trade, dropping precision for speed, is the whole point. It is the technique to reach for when you have a pile of work and need its rough shape in the next ten minutes, not a story-point estimate you would defend in sprint planning.

What t-shirt sizing is

T-shirt sizing is relative estimation with the numbers taken off. Instead of asking “how many points is this?”, the team labels each item XS, S, M, L, or XL, judged against the other items in front of them. An M is an M because it is clearly bigger than that S and clearly smaller than that L, not because it maps to any particular count of hours or points.

That is a feature, not a shortcut. Numbers invite a false argument, the “is this a 3 or a 5?” debate that eats refinement meetings. Sizes are coarse enough that the argument rarely starts: people agree far faster on “that is a Large” than on which specific number a Large is worth. You lose granularity and you gain speed, and early in planning speed is what you actually need.

The catch is the one every relative technique shares. A size only means something if the team shares a reference for it. Agree, out loud, on a reference Small and a reference Large early in the session, or the sizes drift: one person’s M is another person’s L, and by the end the backlog is sorted against five private yardsticks instead of one shared one.

When to use it (and when planning poker is better)

Here is the position, plainly: t-shirt sizing is for shape and speed, not for sprint commitment.

Reach for it when the work is early and rough. A fresh backlog nobody has sized yet. A quarterly roadmap where leadership needs to know whether an initiative is a Small or an Extra Large before anyone writes a story. A first pass across fifty items to find the handful that are clearly too big and need splitting. In all of these you want coverage and momentum, and a coarse size per item is exactly enough signal.

Reach for planning poker instead when the stories are near-term and about to be committed. At that point you are not trying to sort the backlog into piles, you are trying to build shared understanding of a specific story before the team promises to deliver it. The private-vote-then-reveal mechanic surfaces the disagreement that t-shirt sizing smooths over, and near a sprint boundary that disagreement is the most valuable thing in the room. A quiet L hiding a serious unknown is fine on a roadmap and dangerous in a sprint.

So the two techniques are not rivals: they sit at different distances from delivery. T-shirts do the early, wide, rough pass. Planning poker does the late, narrow, careful one.

How to run a t-shirt sizing session

The session is short by design. Five steps:

  1. Set the scale. Decide which sizes you are using (XS to XL is common; some teams use S, M, L only) and, critically, agree on two anchors: a reference Small everyone recognizes and a reference Large everyone recognizes. These two items calibrate the whole session.
  2. Size against the anchors. Take each remaining item and place it relative to the two references. The question is never “how big is this in the abstract?” but “is this closer to our Small or our Large?”
  3. Work fast and let the coarseness help. Do not agonize between neighboring sizes. If an item feels like it is on the S or M line, the fact that it is hard to call means the difference does not matter much yet. Drop it in and move on.
  4. Discuss only the disagreements. Where people land on the same size, there is nothing to talk about: record it and keep moving. Spend the meeting’s oxygen only on the items where sizes genuinely diverge, because that gap is a real difference in understanding worth surfacing.
  5. Flag the XLs for splitting. Anything that lands on XL is usually a signal, not an estimate. It is too big to plan confidently and is a candidate to split into smaller stories before it goes anywhere near a sprint.

Turning sizes into story points

This is the bridge teams get stuck on, and it is worth being precise about. At some point the early, rough sizes have to become something you can plan a sprint against, and a plain mapping is the most useful thing you can give the team.

T-shirt sizes mapped to story points A rough band, not a fixed formula XS S M L XL 1 2 3-5 8 13+ rough story points L and XL are usually a signal to split, not a number to commit
A mapping gets you into the right neighborhood. Planning poker settles the exact number.
T-shirt sizeRough story pointsWhat it usually means
XS1Trivial and well understood, done in an hour or two
S2Small and clear, no real unknowns
M3 to 5A normal story with a few moving parts
L8Big or uncertain, a candidate to split
XL13 or moreToo large to commit, split before sprint planning

Two warnings come with that table. First, it is a starting band, not a fixed formula. A t-shirt size is deliberately coarse, so the same L can honestly land on a 5 or an 8 once you look closely; the mapping gets you into the right neighborhood, and planning poker settles the exact number. Second, and this is the one that bites: do not use the mapping to skip the re-estimate. The sizes were a fast gut-check across the whole backlog. For any story about to enter a sprint, re-estimate it with story points properly, using the size only as a sanity check. If the size said L and the team votes it a 2, that mismatch is a conversation, not a rounding error.

Because the sizes do not add up (an S plus an M is not an L), you cannot sum them into a velocity number. That is the hard boundary of the technique: t-shirts give you shape, story points give you a number you can measure velocity from. Keep the two jobs separate and both stay honest.

A worked example, start to finish

Say a team sits down with a fresh backlog of six items and ten minutes.

They pick two anchors first: “update the copyright year in the footer” is the reference Small, and “rebuild the checkout flow” is the reference Large. Everything else gets sized against those two.

  • Add a tooltip to the settings icon: clearly near the footer change. XS.
  • Add email validation to the signup form: small and well understood, a touch more than the tooltip. S.
  • Build a CSV export for reports: a normal story, a few moving parts, no scary unknowns. M.
  • Add SSO via a new identity provider: big, and nobody is sure how the provider’s API behaves. Two people say L, one says XL. That disagreement gets a two-minute discussion, which surfaces the real question (is this one provider or a general SSO framework?) and they settle on L, with a note to split it.
  • Migrate the reporting database: everyone independently lands on XL, so there is nothing to discuss except that it is too big and needs splitting before anyone commits to it.
  • Fix a typo in an error message: XS, no discussion.

Six items, sized in a few minutes, with the two genuinely uncertain ones flagged for a closer look. The team now knows the shape of the backlog: mostly small, one large item to break down, one extra-large that is really an epic wearing a story’s clothes. Next week, when the top of that backlog is refined and heading into a sprint, they will re-estimate those near-term items with planning poker and let velocity build from the points. For a wider view of where t-shirt sizing sits among the alternatives, see the estimation techniques comparison.

Estimate together in TeamRetro

T-shirt sizing works best when everyone sizes at once and you can see the disagreements at a glance. TeamRetro’s estimation tool runs the whole session in one place: size items as a team, reveal together, and discuss only where you diverge, then switch to free planning poker with story points when the near-term stories are ready to commit. Shape early, commit late, in one shared board.

Frequently asked questions

What is t-shirt sizing in agile?

T-shirt sizing is a relative estimation technique where the team sizes each piece of work as XS, S, M, L, or XL instead of assigning a number. It is a fast gut-check of how big something feels compared to the work around it, not a measurement of hours or story points. Because the sizes are coarse, a session moves quickly, which makes it a good fit for early refinement and roadmap planning where you need rough shape, not precision.

How do you convert t-shirt sizes to story points?

Agree a simple mapping as a team, for example XS to 1, S to 2, M to 3 or 5, L to 8, and XL to 13 or more, then re-estimate the near-term stories with planning poker rather than trusting the conversion. Treat the mapping as a starting point, not a fixed formula: a t-shirt size is a rough band, so the same L can land on 5 or 8 depending on the story. The conversion is for turning a sized backlog into something you can plan against, not for producing a final number you commit a sprint to.

When should you use t-shirt sizing instead of planning poker?

Use t-shirt sizing when you need to size a lot of work quickly and roughly: a fresh backlog, a quarterly roadmap, or a first pass before anything is refined. Use planning poker when the stories are near-term and about to enter a sprint, because that is when you need the shared understanding and the surfaced disagreement that a private-vote-then-reveal discussion produces. T-shirts are for shape and speed; planning poker is for the commitment.

Can you measure velocity with t-shirt sizes?

No. Velocity is an average of story points completed per sprint, and t-shirt sizes are categories, not numbers: an S plus an M does not equal an L, so the sizes cannot be summed into a per-sprint total. If you build a conversion table to force them to add up, you have reinvented story points with an extra step. Size early work with t-shirts for shape, then re-estimate the near-term stories with points and measure velocity from those.