Scrum teams sometimes use t-shirt sizes to avoid over-specific estimates. Choosing between small, medium and large allows a rough initial product backlog prioritisation. T-shirt sizes sound simple, but they are as much of a recipe for misunderstanding as hours estimates.
Mike Cohn explains that there are two problems with t-shirt size estimates, one of which is that ‘your view of an XL may not match mine’. The problem is that a word like ‘large’ can lead to misplaced arguments like ‘that isn’t large’ when the sizes were only supposed to be relative to each other.
Dress size estimates
When I met AntonyMarcano, and t-shirt size estimates came up, he suggested using dress sizes instead. Dress sizes give the illusion of numerical objectivity, but any woman knows that there is no standard size 10, and that the actual size varies from shop to shop. Anthony even claimed that the actual measurements depend on the time of year. In fact, dress sizes are so problematic that Antony may not have been entirely serious.
Dress size estimates are more like story points - a relative numerical scale that everyone knows to take with a pinch of salt. In practice, dress size estimates make a good conversation piece but are likely to be trouble on an actual project: you’d do better with a more obviously subjective relative scale. To find a better analogy, let’s turn from retail clothing to restaurant menus.
Chilli rating estimates
Restaurant menus sometimes indicate spicy dishes with a chilli rating - a certain number of chilli symbols. The scale is variable and often not explicit: you might think that there are zero to three chillies, until you turn the page and find the single dish with a four chilli rating.
What’s clearer with chilli ratings is that one restaurant’s scale is unrelated to another’s. One region’s cuisine is hotter than another’s. As with dress sizes, the scale also varies from country to country to accommodate local tastes; three chilli symbols in an Indian restaurant in Paris doesn’t amount to much.
Perhaps your Scrum project doesn’t need chilli ratings, but if you find yourself using sizes that have too much baggage, changing to a new unit might be a useful way to reset expectations about how precise these estimates are. After all, with estimation, you sometimes need a way to express that are not precise more than you need to make your estimates precise.