Estimating software development. Pain.

At work today, I had the wonderful opportunity to spend an hour handing over guardianship of the Estimate sheet. I used to be the sad character who would update that, for any and every misguided idea that anyone in the company would come up with to finally get estimating right. I suppose my disbelief – which has only grown in the eighteen months of dragging this artefact behind me – and lack of enthusiasm started to show.

From initially being ambivalent and doubtful about the value of estimates I have now become utterly convinced of their uselessness. I have stopped asking for estimates from my teams, and I have stopped giving them to people who say they need them. So far that is working just fine.

This presentation (https://youtu.be/QVBlnCTu9Ms) from the #noestimates movement – yes there is such a thing – sums it up very well. It makes me smile.

Allen explains why estimates hurt so much. So what do I do now? The only thing that really works. I skip estimate sessions, and just keep a prognosis by applying linear maths on observed progress. My focus for teams is to help them get into a sustainable pace. Then we measure what that pace is to calculate a prognosis of when our current workload has room for a new thing. We don’t interrupt ourselves either with urgent tasks. One thing at a time, but small enough tasks that every day we finish something useful.