Rys Sommefeldt

Accurate project estimates

I’m doing project management training at work and one of the things brought up by the external trainer was the requirement to be completely honest about the initial cost (time, money, etc) estimates, even if that means saying you don’t know. The basic idea, which is such obvious common sense that it barely needs saying, is that there’s no point estimating at all if the margin for error is high.

Rather than put any time or money estimate down, be up front about needing more data first and estimate accurately on that work instead. Often in the first stages of a project you can only give a good estimate on the cost of a body of work after working on a smaller part of it first.

The reason I even write about it is I was nearly fired for refusing to provide concrete estimates at a previous job, on work I had no idea how to reason about. I was effectively coerced into lying about costs just so the clueless project manager could give data to senior management and look like he was in control, and so I wouldn’t get thrown off the project or worse.

In the end, after delivery, I got a formal reprimand and dressing down for not wanting to iterate on the work for a week first so I could be accurate. I decided there and then to quit. The best bit is the project was ’early’ because I purposely went high on the demanded estimates, because I didn’t want to put myself under pressure over something I had no good idea about.

It’s refreshing that Imagination are happy for their engineers to be as accurate as possible during project estimation, even if that means an unknown amount of work needs to be done first. Engineering is in the business of truth after all.