Ranges and probabilities

Ranges help me, and my clients make better decisions. I’ll always say that I expect between X and Y as a result. My colleague always says he expects Z as a result. I’m usually accurate; my colleagues precise forecast is generally wrong.

Precision is how close we can get numbers to match each other. We focus more on details.

Accuracy is how close we can get to the actual result. We focus more on making sure we get it right.

A very detailed model that takes into account hundreds of signals gives you a specific estimate. Still, it tends not to be very accurate. Accuracy lets you make better decisions than you can with a precise model.

Precision makes you feel good because of all the details.

It makes you feel like you’re in control because you’ve got a specific number to back you up. But there’s an issue with precision because it leaves no room for unknown variables. It gives the illusion of certainty. We will get around 129 widgets. But what does around mean for each person? What if we miss a crucial variable in our model? You get very precise and end up being wrong.

Accuracy leaves the door open for the unknown.

We’re not getting a specific number but a range. We might get 130 widgets; we might get only 115. There’s room to play with here. We can have a conversation about it now, and we can still steer the ship in the correct direction.

Getting a wrong but precise estimate most of the time puts your relationship with your stakeholders under stress. Overall it’s better to be more simplistic in your calculation and accurate, than precise and wrong.

Superforecasting & thinking in bets.

It’s what superforecasters do. They leave room for what they don’t know by working with ranges. Then they update their forecasts based on new information. That means when they start, they aren’t precise but still tend to be accurate. They adapt and change.

Consider Thinking in Bets which underlines the need to embrace ambiguity in decision making. It makes the case that we don’t know everything. We need to communicate confidence levels. Instead of saying that we need one specific course of action, we say there are two but add probabilities to each one. “I’m 80% sure we’ll sell 120-130 widgets”.

Communicating in ranges and probabilities takes a little getting used to. As consultants or experts, we’re looked upon to provide the answer. So we puff up our best guess to be just that. But what if we’re wrong? What if we need other options?

That’s why ranges and probabilities allow us to present alternatives, with an estimated impact (ranges) and ranked by how effective we expect them to be (probabilities).

Leave a Reply

Your email address will not be published. Required fields are marked *