How to estimate user stories: the 12 Spokes rubric
We held a company retreat in a lodge at Sundance Resort a few weeks ago. We made fantastic pizza and played cutthroat ping pong, but we also spent a great deal of time discussing our process. I’ll post some of the highlights in the next few weeks. We spent a session discussing how we estimate user stories to make sure our velocity in Pivotal Tracker is as accurate as possible. We figured it’d be a helpful thing to share.
First, you need to know that we use the “power of two” estimating system. Rather than using the standard 0,1,2,3-point system for designating low-effort to high-effort stories, we feel like it’s more accurate to use a 0,1,2,4,8-point system. An eight-point story is obviously much larger than a one-point story; at least, it’s more obvious than the difference between a three-point story and a one-point story. That said, here’s our rubric.
- O points: There are absolutely no unknowns; there’s no question it’s going to work. Examples include copy changes and color changes. If you have to write a spec, it’s not a zero point story.
- 1 point: Things that require the simplest unit of real work. There’s pretty much nothing unknown about it, doesn’t require new architecture; it’s a permutation to a well-specced existing feature. It’s isolated; it touches one part of the app and isn’t interlaced throughout. One-point stories can include items that need some design review.
- 2 points: Stories that are bigger than this can generally be broken down into smaller pieces. Two-pointers are within the realm of “I know how to do this.” You know which tool you’ll use. It touches the database, requires testing; there may be some uncertainty, but you know what fixes this problem. While we don’t like to put time limits on stories, a two point story is generally going to take your morning or your afternoon.
- 4 points: These are rare. These are the stories you don’t know how to break down any smaller because you don’t know the right approaches. These are the stories you preface with the response, “Well….” To put it in time-specific terms, these stories take days or multiple days.
- 8 points: “Oh shit. Well…..” These are stories you know are possible but you have no idea about how it’s going to happen. It’s something we need to do research on and plan. These are stories that will take half a week, a week, or more.
Dec 02, 2011 @ 17:06:44
IMO, there should be no such thing as a zero point story. anything that needs to be done, that isnt already done, take some amount if effort (1 point). this is important because you want to measure this something, even if it’s very little, but more than nothing. also, i tend to think of story point size as a measure of complexity, not just how uncertain you are. you might create an epic that is 8 points due to its complexity (and uncertainty). of course, the larger stories need to be decomposed into smaller stories fit for consumption. it is valuable tho, to use the larger stories for epic planning. then, at the latest responsible moment, decompose the epics into consumable stories (down to the smallest testable unit of business value).
Dec 02, 2011 @ 17:22:00
i meant “effort”, not complexity. Story points are an approximation of effort. the amount of effort is affected by complexity & uncertainty.