Beating Estimates Should Be a Punishable Offense

Friday, December 19, 2014

I'm completely sick of hearing how individuals, teams and companies beat their estimates. It is not something to be celebrated. It is a business anti-pattern that should draw the ire of managers everywhere.

Am I a scrooge for saying this? Maybe, but I have folks' best interests at heart. Before I go further, I have to ask. What is an estimate?

estimation with a gantt chart
Image courtesy of Wikipedia

An estimate is nothing more than a prediction about the future. I could estimate how long it will take me to write my next book, but the actual time required will be highly dependent upon many facts that I do not yet have. Before work commences, I might not know the intended audience, the amount of research required, the competition - even the topic. I won't even know if consulting work will get in the way of my schedule or if I will get sick with the flu. Because of all of these unknowns, my estimation is subject to a significant amount of error - especially toward the early stages of a project.

Yet, somehow, people forget that estimates are created when a project has many unknowns. These poor souls subconsciously promote their educated guesses to actual constraints upon their projects. To not live up them becomes tantamount to failure. This is a bad thing!

As a project manager, I worked with one fellow who proudly proclaimed that he always beat his estimates - he always finished both under-budget and under-schedule on every project he led.

Though he hoped that his boasts would impress me, I just shook my head. His proclamation was nothing more than an admission that his estimates were useless and bore little resemblance to reality. A great estimation will be just as likely to err too high as too low. A series of estimates that were beaten on a consistent basis is no mark of honor on a team, so much as a mark of failure upon the estimator.

Yet the constant political pressure to beat expectations and estimates is such that all estimators are forced to overestimate. To produce actual estimates would be tantamount to risking one's livelihood.

So what's the solution? I propose a new term: "failure-prone estimates." The term seems scary, and as such will carry no expectation that anyone meet or exceed them on any given effort. Sometimes a new name is they only means of bringing back an older term whose name has been subverted.

If you'd like to learn all about estimating pricing power for software systems, you should check out my book - either that or contact me for a private consultation.