You know what I'm talking about (1 x the worst) + (1 x the best) + (4 x the most likely) all over 6. I guess that not any people can think it's that great, because I've never encountered anyone using it.
What then is the problem? Well first, what is the level of confidence in the output? Dunno. And I mean, really, you don't know. The principal problem is you don't know where the input figures came from. Likely as not, a bit of a stab in the dark in a workshop. At best, it will be sourced from experienced team members who have done something very much like the work before. This doesn't guarantee much as we shall see, but it's an improvement. Oh and the best bit, you're just as likely to corrupt a good figure as average out a bad one.
So what's wrong with a stab in dark? It has its place surely? We can refine things as we go can't we?
Let me cite two quick examples.
I'm working with a colleague over the phone - he's got the job of assembling a project plan. He's asking me about specific aspects of my work and putting 'durations' next to the work. We talk about a particular task and he offers that 2 days seems reasonable. I counter with 1 hour. Quite correctly he challenges this disparity. I'm able to have some level of confidence in my figure because I'd already done the work - and it took about an hour. We then quickly move towards a second activity which utilises the output from the first task. My colleague offers that the work will take around three weeks. I counter with 45 seconds. What he thought was a manual task was a bit of cut and paste with a spread sheet.
If Pert weighted estimating is used in these two cases then the first example will be off by around 800% and as for the second, I'm not even going to waste time with a calculator.
So, two quick examples of how things can go off the rails. Now someone out there will be saying hold on, hold on; Pert weighted averages are only as good as the data that goes into them - same as any forecast. I agree up to a point, but there is no discussion or reference anywhere to this in the literature. Is there? Moreover, where do you get an optimistic estimate and a pessimistic estimate from? I've only got my best estimate, but usefully I'll tell you where it came from and often, what my level of confidence is in the figure. Oh - and not forgetting that there's not mention of how to deal with estimates that are either highly consistent or wildly divergent - and these two scenarios should be treated differently shouldn't they?
Let's set out a few immutable truths. Or at least some points which I hope won't raise too much debate.
Good things arising from good forecasting
- Better project plans that build confidence
- Better stakeholder engagement and perception - they know what's coming, when and in what order
- Better management of commercials - you can sign those distracting multi-million pound contracts with a little less anxiety
- Better expenditure profiling - you know what costs are coming when
- Better resource planning, and importantly brokering. Just 'cos I said you could have 2 engineers this week doesn't mean you can have 'em next week...
- More productive project boards - you're not arguing about the schedule all the time
- A happier project sponsor
...and quite a bit more besides.
Bad things arising from bad forecasting
Quite apart from the absence of points 1 - 7 above are the following.
- You'll constantly be re-working the project plan and your monthly project board will become your monthly re-baselining meeting.
- Your project will quickly become a source of generalised uncertainty, consternation and (quite possibly) resentment
- Your chances of getting to the finish line and fulfilling anyone's definition of a successful project will fairly quickly diminish to nil
- And a real kicker here, you'll dilute your governance (who makes what decisions, when and to what criteria)
As is so often the case, my blog post has ended up in quite a different place than I anticipated (that of course is attributable to poor forecasting) but between this article and those yet to be published I'll try to provide adequate coverage of the following.
- Parametric or reference class forecasting
- Guessing, uncertainty and pragmatism
- Never mind 6 Sigma - 1 will do you quite nicely
- The PMs job in fighting for exactitude and rigour in the planning process
- Some helpful language and strategies to challenge sub-standard practice
- Some real life scenarios and tools consistent with other articles in this blog that I hope will add a bit of value