T-Shirt Sizing

Copione di facilitazione

1. 1

   Open by stating what the sizes measure: relative effort and complexity, not hours or deadlines. Show the scale from XS to XL and agree whether XXL is allowed — anything that big usually needs splitting anyway.

   5 min
2. 2

   Calibrate anchors: ask the team to nominate one recently completed piece of work everyone agrees was a clear M and another that was a clear S. Write both where they stay visible all session — every estimate is relative to these two.

   10 min
3. 3

   Run a practice round on a throwaway item: silent pick, simultaneous reveal on a count of three, and a short outlier discussion — so the mechanics are settled before the real backlog starts.

   5 min
4. 4

   Size the real items one at a time: the item's owner explains it in under a minute, everyone picks silently and reveals together. On consensus, record the size and move on; on a spread, give the highest and lowest voters one minute each to explain their reasoning, then re-vote once.

   30 min
5. 5

   Park any item that still splits the room after a re-vote — repeated disagreement means it needs splitting or research, not a longer argument. Review the parking lot and assign a follow-up owner per item.

   5 min
6. 6

   Close by reading the full list back with its sizes, ask whether anything now looks wrong in context, and state what happens next — for example, L and XL items go to refinement for splitting before any planning.

   5 min

Suggerimenti

• Use T-Shirt Sizing for epics and features. Use Planning Poker for sprint-level stories.

• Disagreement is signal — two people with very different sizes have different mental models. That's worth a 2-minute discussion.

• Don't let the session become about hours. It's about relative complexity.

• Keep a 'parking lot' for items that generate repeated disagreement — they need splitting.

Errori comuni

• Skipping anchor calibration — without a shared reference M, every size comes from a private scale and the results are incomparable

• Letting the discussion drift into hours — the moment someone says 'an M is about two weeks', the method collapses back into the false precision it was meant to avoid

• Voting sequentially instead of simultaneously, so the first confident voice anchors everyone who follows

• Debating an item past the second vote — persistent disagreement is a signal the item is misunderstood or too big, not a debate to be won

Variazioni

Bucket System: place items in buckets (XS, S, M, L, XL) rather than estimating one by one — much faster for large backlogs (50+ items). Animal Sizing: some teams use animals (mouse=XS, elephant=XL) to further remove numeric anchoring.

Metodi correlati