If you read my recent note on Estimating With Time And Relative Sizes you might be interested in this method of generating relative sizes.
Firstly, some important ground rules. You want a group of people to generate a size simultaneously and you don’t want to let anyone say numbers out loud, or undermine another individuals thought process before the moment of sizing. Phrases like “seems pretty simple” or “can’t be that hard” or “clearly this is a 3, but what does everyone thing?” are all totally unacceptable. Everyone should make their own minds up and nobody should ever be scared to give a radically different answer – what did they think of that nobody else did?
Once sizes are shared, nobody should be teased, punished or made to feel stupid about the number they gave. Discussing the differences is good. Creating an environment where everyone just wants to give the same answer as everyone else to avoid pain is bad.
So in a lot of teams, the moment of sizing involves holding up a playing card with an appropriate number on, or holding up fingers, or some other method to reveal the choice at the same time. This is all fine.
The gap in most processes is the lack of references. If you are truly trying to create relative sizes, it helps a great deal to have examples to hang the decisions off of. If you hold up a t-shirt and ask people what size it is, not many people will guess the right answer. If you can only see a “large” tee, you could easily guess “medium” or “extra-large” because you have nothing to compare it to. If you held up examples of previous small, medium, large and extra-large t-shirts it is suddenly much easier to gauge the size of the new t-shirt compared to your reference t-shirts.
So don’t go into a planning meeting without examples of previous stories that are well known and considered representative of the groups of sizes you use.