The allure of #NoEstimates

 

A project manager asked me a question which I’ve frequently been posed over my career.

How should I deal with a stakeholder who, when I provide a rough order of magnitude ranged estimate early in the life of a project, insists on holding me accountable to the lower end of that range later on even when sufficient evidence has emerged to contradict that value?

The question clearly states that a range of values was provided and no commitment was made. And yet, the behavior of the stakeholder was the same as if a single point, firm fixed estimate had been given.

Anchoring and confirmation biases can partially explain why this happens, but this provides little assistance to a project manager who is asked to provide an estimate at the beginning of a project.

But even if more than 50% of the project scope has been delivered, we might still be unable to provide an accurate estimate.

We learn that a simple way of calculating Estimate At Completion for a project is to base it on past performance but how often is that the case in reality? To provide an accurate forecast, we would need a delivery process which is in control, and yet, most of the time we may have limited influence over factors which could cause a predictive model to break down.

If we don’t have reliable availability of people, regardless of which delivery approach we use, we won’t be able to predict when we will be finished until all scope has been delivered. Metrics such as velocity, throughput or average work item age are as helpful as a magic eight ball under such conditions. Ensuring that everyone is available when needed is not impossible, but it is difficult to achieve when delivery commitments are made without taking an organization’s ability to deliver into consideration.

But even with dedicated staffing, unless the level of uncertainty associated with the remaining work is less than or equal to what the team has experienced to date, past history won’t be indicative of future performance. Risk management helps by encouraging a team to address high severity threats as early as possible, but that assumes that they are able to identify key threats. The more complex a project, the greater the difficulty in doing so. All it takes is the realization of one particularly nasty unknown-unknown to invalidate a high confidence estimate. Contingency and management reserves provide a degree of shock absorption but on extremely complex projects, the tail of the poor outcomes is long indeed.

A container ship got stuck sideways in the Suez Canal this Tuesday. It is not the first time such a shipping issue has occurred yet four days later no one is able to provide an accurate estimate as to how quickly the the canal will be unblocked.

If you are managing a project which is unlike any other, why would you expect to be able to do any better at forecasting when you will be done? #NoEstimates might not be acceptable to many stakeholders, but it might be the most responsible answer in some circumstances.