T-shirt sizing for agile estimation

T-shirt sizing is a relative estimation technique where a team labels work with clothing sizes, usually XS, S, M, L, XL and XXL, instead of numbers. Like story points, the sizes compare items to each other rather than predicting hours. Because there are no numbers, people focus on the size of the work and not on arithmetic, which makes t-shirt size estimation in agile teams quick and approachable.

How t-shirt sizing works

The team agrees on what each size roughly means, then places backlog items into those sizes. A typical scale:

Size	Rough meaning	Example
XS	Trivial, well understood	Fix a typo in a confirmation email
S	Small, few unknowns	Let users change their display name
M	Moderate, a few moving parts	Add password reset by email
L	Large, touches several areas	Add two-factor authentication
XL	Very large, significant unknowns	Support login with third-party identity providers
XXL	Too big to reason about as one item	Rebuild the account system

The examples are only anchors. Your team should pick its own reference items from work it has already done, the same way it would choose reference stories for numeric estimates.

You can size items with a quick vote (everyone picks a size, reveal together, discuss outliers, as in planning poker), or by sorting cards on a wall or virtual board into size columns.

When t-shirt sizing beats numbers

Roadmaps and early planning

When you are deciding what to build next quarter, you rarely need more than "small, medium or huge". T-shirt sizes give that answer without pretending to a precision you do not have.

Large, early backlogs

If you have 80 rough ideas to triage, discussing each one at Fibonacci precision would take hours. Sizing them with t-shirts can be done in a single session, and only the items that make the cut get refined and estimated in more detail later.

Mixed audiences

Stakeholders understand "this is an XL" immediately. Numbers like 13 tend to be read as days or hours, no matter how often you explain otherwise.

New teams

Teams that have never used relative estimation often find sizes easier to start with. Once they are comfortable comparing work, moving to story points is a smaller step.

When numbers work better

T-shirt sizes cannot be added up. "Three Ms and an L" does not tell you whether the sprint is full. For sprint planning and velocity-based forecasting, most teams switch to story points on a Fibonacci scale once stories are refined.

Converting t-shirt sizes to story points, carefully

Sometimes you need numbers, for example to get a rough total for a release. A common approach is to map each size to a point value:

• XS = 1
• S = 2
• M = 3 or 5
• L = 8
• XL = 13
• XXL = too big, split before converting

There is no standard mapping, so agree on one as a team and write it down. Keep these points in mind:

• The result is a rough total, not a plan. A release estimate built from t-shirt conversions has wide error bars. Communicate it as a range.
• Re-estimate before committing. When an item moves into a sprint, estimate it again in story points with the details you now have. Do not just copy the converted number.
• Don't convert XXL. An XXL means "we don't understand this yet". Turning it into 40 or 100 points hides that.

Pitfalls to avoid

Too many sizes

Adding XXXS or XXXL brings back the false precision t-shirt sizing is meant to avoid. Five or six sizes are enough.

Undefined sizes

If each person has a different idea of what an M is, the sizes mean nothing. Anchor each size to a real, completed example.

Sizes that quietly become time

"An M is about a week" is tempting, especially for stakeholders. Once that mapping exists, people start estimating in weeks again. Keep sizes relative.

Never refining

T-shirt sizes are a first pass. If L and XL items sit in the backlog for months without being split, the sizing exercise has not done its job.

A worked example

A product team is planning the next quarter and has a list of account-related ideas. They agree that "Let users change their display name" is an S and "Add password reset by email" is an M, because both were built recently and everyone remembers them.

The first new item is "Add two-factor authentication". Votes come back as two Ls, one M and one XL. The developer who voted M assumes an off-the-shelf library will do most of the work; the one who voted XL points out that recovery codes and a support process for locked-out users are also needed. After a short discussion the team agrees on L and notes that recovery could become its own item. The whole exchange takes about three minutes, which is the point: enough conversation to catch a missing piece, not enough to design the feature.

Running a t-shirt sizing session

1. Prepare the list of items, ideally with a one-line description each.
2. Agree on the reference examples for S, M and L before you start.
3. For each item, everyone votes privately and the votes are revealed together.
4. If votes span more than one size apart (for example S and XL), ask the two outliers to explain, then vote again.
5. Record the size and move on. Aim for a minute or two per item.

Scrum Poker includes a t-shirt sizes deck, so you can run this online: paste the items into the agenda one per line, share the invite link, and the session steps through them. The voting timer (30 seconds, 1 minute or 2 minutes) helps keep a large backlog moving. For a side-by-side look at t-shirt sizing and other methods, see agile estimation techniques compared.