Foreseeing the Future

As a new year starts, it seems that (nearly?) all journalists are busy making their forecasts for the future. Of course, at the end of the year, we may be given the opportunity to look back at their predictions and see how wrong they are. They are currently predicting that President Trump will not last the year, or that he will last but not get reelected next year, or that he will get reelected, or that he will be in prison, or that he will stand down in exchange of amnesty, or just about anything else. As we listen to these predictions, we must think to ourselves that these journalists don’t really know or believe themselves what they are saying – it is mostly making predictions based on personal desires or fears…

And yet, in business, we believe that all estimates need to be contractually binding, transformed into deadlines, with penalties for “non compliance” should a delivery be a day late. I have already written about estimating problems, so I will not return to that topic.

However, I do want to remind you that an estimate, ultimately, is just guesswork. Perhaps you have a lot of data to demonstrate that, in the past, this was how long something took; perhaps you have done a statistical analysis and can show root causes of variations, but it is still, in the end, just a guess.

When requesting an estimate from someone, you should always ask for a confidence level. I tend to recommend an estimate to be provided in the form of an S-curve:

S-Curve
Probability and deviation in estimating

This approach allows you to say that you will only feel 100% comfortable if you are given at least x (in this picture, 703 days) to complete the project. If you want to negotiate down to 500 days, I feel 80% comfortable that we can deliver; if you are willing to take the risk, we can work together.

A consequence of this is that, as I have frequently said:

  1. an estimate should never be translated into a deadline;
  2. always make sure that you are clear why you are asking this and what you plan on doing with that estimate.

Language is tricky. If you request an estimate, a forecast or a prognosis, you may get very different data depending on how people interpret that word. If you request a projection, an evaluation, an assessment or an opinion, who knows what you might get in return.

But remember, ultimately, it’s just a guess.