Estimation Scales
T-shirt sizing replaces numbers with sizes like XS, S, M, L, and XL. It removes the pressure to be precise and lets teams sort work by rough magnitude quickly, which makes it ideal for early backlog triage and cross-team roadmap planning.
Numbers invite debate about exact values. Sizes invite comparison. Asking whether a story is a medium or a large is a simpler question than arguing between a five and an eight, so conversations stay quick and decisive.
Sizes also communicate well to non-technical stakeholders. A product manager or executive immediately understands that an extra-large item carries more risk and cost than a small one, without needing a points primer.
Present each item, let the team ask questions, and have everyone pick a size privately. Reveal at once and discuss any wide gaps, just as you would in planning poker with numbers.
Many teams keep a reference item for each size so estimates stay consistent over time. A shared example of a medium story anchors the whole scale and reduces drift between sessions.
T-shirt sizing shines for high-level planning, but it does not aggregate into velocity easily. Once a team needs to forecast how much fits in a sprint, mapping sizes to a point range or switching to a numeric scale gives sharper data.
A practical hybrid is to size the whole backlog first, then estimate the top items in points as they approach a sprint. Sizing sets priorities; points support commitment.
Most teams use four or five sizes. Fewer than three loses useful detail, and more than six recreates the endless debate that sizing is meant to avoid.
Yes. Teams often map sizes to point ranges, such as small to two or three and large to thirteen, once they want velocity data.
It is most valuable early, but some teams keep using it throughout if they prefer speed over granular forecasting.