Agile Estimating From The Beginning

Hi, I'm a software developer working an agile project and I came into this project in the middle of it, so I never got to see what happened (such as scheduling) at the beginning of the project. Anyway, I've been doing a lot of reading on creating a product backlog using story points, and one of my questions is, how do we translate that to actual time without going into too much detail up front? Inevitably, at the beginning of the project (or even if I am writing a proposal for a new project), someone will ask, "How long will this take?" I understand that it's impossible to give an accurate estimate that early on, but I have to give some estimate? Should I start trying to estimate what our velocity will be based on previous projects? And what if it's a new project? Any thoughts?

Hello,

I think this is a complex topic. The best book I know on the topic is the one from Mike Cohn "Agile Estimating and Planning" (http://www.mountaingoatsoftware.com/book/1-agile-estimating-and-planning)

John

John,

Thanks for that thought. I had actually already ordered that book and I'm getting ready to read it. Thanks!

Jeff

I know of three basic approaches to the problem:

- estimate in ideal time and apply an arbitrary factor between e and pi (for example 3)

- estimate in whatever unit you are used to and use past experience with similar projects and/or teams to guess how that translates to real project time

- after estimating the project, develop two or three representative features, and use that experience to extrapolate the duration of the whole project

Especially the first approach works stupendously well.

Thanks Ilja. You'd think by now us developers would multiply our own estimates by 3 so we can estimate correctly. If only we weren't so optimistic...