Beforecasting
What is going on when we forecast something? We know we can’t see into the future, and yet we try. We predict. Think of all sorts of estimates at work, for example. They are predictions based on our understanding of the circumstances, our skills and our experience. And, presumably, logic.
It is one thing to apply logic in retrospect for insight, and another as foresight. You might remember a question posed by a well-known fictional character: How often have I said to you that when you have eliminated the impossible, whatever remains, however improbable, must be the truth?
There’s an awful lot hidden in that question … the ideas of possibility, probability, truth. Sherlock Holmes’ deductive methods are famous the world over and have inspired many a detective spinoff. The practice of root cause analysis has a shared heritage. But these methods look at the past. What tools do we have to look at the future?
How can we know what will happen? If we’re honest with ourselves, the ultimate answer must be “We can’t.” Instead, we need to embrace uncertainty and alternatives. Sure, if you predict that the sun will rise again tomorrow, you are extremely likely to be right. There’s a high degree of confidence in such a prediction and little uncertainty. It is an example of cases for which I’ve devised my own sniglet. I use the word beforecasting to describe making predictions based on things that have happened many times before (whether one recognizes the underlying cause-and-effect mechanisms at play, or not). There are many aspects of product development that can benefit from beforecasting. Many more simply can’t.
Predicting whether your solution will sell, or even if it will work, is very different. It is a try-and-see problem. Logic and analysis alone are insufficient. They apply but need to be supplemented by robust experimentation, measurement, statistics and, if possible, simulation. It’s less about predicting and more about managing investments in the acquisition of knowledge. The only reliable forecast you can make in such situations is that you will learn. To bend your brain a little, I’d like to make a beforecast: Understanding what there is to learn and how much you’re willing to risk to do so is not a job for beforecasting.
Consider
How often do you succumb to the temptation of applying beforecasting when it simply isn’t appropriate?
Additional Insights
A probabilistic perspective
Use your data to be more agile
Intuition versus data
