Estimation: Why do we get it wrong so often?
Posted by Isabella Schembri on 3rd Jul 2017
The East of England branch event Estimation: Why do we get it wrong so often? and AGM took place at University of Hertfordshire, Hatfield, Thursday 22nd June 2017. The talk was given by Charles Willbe, an accredited MPA High Risk Gateway Review Team Leader.
He is currently a partner in Project Doctors and beforehand he worked for GlaxoSmithKline for 25 years in various roles.
Estimation is difficult for the simple reason that projects are unique. Underestimation is common whereas overestimation is very rare and the bad news is that we are not getting any better at estimating accurately. When thinking about producing accurate estimates, we must be estimating within a range and not as a point, if we want a more realistic approach.
Charles challenged the nearly 20 attendees with a number of exercises on estimation, gradually adding to each of them new data to support the calculation of the estimate. We learnt some interesting facts out of these exercises. One, that overspend and underspend are not symmetric, another that we slow down when we know we have extra time, moreover, that we don’t make up time on the easiest activities, etc… We also unthinkingly provide estimates which must fail 50% of the time and Charles suggested that both we and clients would be far happier accepting no more than a 5% risk of failure, so our estimates must be prepared on this basis.
Breaking estimates into small chunks provides many advantages. Can you envisage what can be achieved in 200 days? Don’t you have a far better idea of what can be done in 20? Or in 2? The shorter the period for the constituent parts, the more accurate the estimate for the whole. Charles was clearly promoting closer deadlines for specific activities to reduce the “student effect” on meeting estimates, i.e. submission to assignments right at the end. He suggests that ideally activities’ lengths should be kept down to 10 days maximum to keep staff focused on meeting the deadline.
Statistical summation of activities to provide better overall estimation was explained and Charles showed the audience how this could be achieved through some examples.
Finally, Charles suggested that the range of times for each activity should be associated with the risks to their completion and opportunities to do them more quickly, in order to provide more accurate and manageable estimates. As well as making our “best guess”, we must look at best case, with opportunities for saving effort, and worst case, when risks become issues.
In summary, this event caught the audience’s attention for being quite unique in the way it tackled poor estimation and, I am sure, it will make an impact in how we think of estimates in future projects.
If you wish to learn more about Project Doctors, please see the contact details in the slide-pack.
The branch AGM took place prior to the session and following nominees were unanimously elected to serve as the branch committee over the coming 12 months:
Mark Buckton, Brent Crossman, Chris MacLeod, Lynn Newman, Charles Willbe, Maria Duncan