Forecasting, Scenarios, and Targets

A Forecast uses a statistical method and historical data to make a statement about what is likely to occur in the future.

A Scenario uses a generalized model, in the absence of contextual historical data, to make a statement about what is likely to occur in the future.

A Target is a statement about what what the future should be.

The linkages between forecasts, scenarios, and target setting are subject to all sorts of phenomenon. Anchor-and-Adjust, optimistic thinking, convenient reasoning, and prospection error all come into play. The gap between a target and the predicted future is either a source for dissatisfaction or for celebration. One uses a forecast to minimize that error, and, ideally, to be smarter going in.

Analytics is involved in understanding the drivers of performance, which drivers are likely controllable or maneuverable, and then projecting that outwards. The experienced scientist avoids overloading. That is to say, people try to endogenize the unmeasurable. This quest for ‘completeness’ may be noble, but it causes the forecast to become less predictive, and, sometimes, unverifiable. The most concrete example I can think of is assuming that content will be relevant and divergent at a much higher degree than it has historically. Introducing a ‘jump function’ into the mix doesn’t help.

My preferred method of forecasting is actually called backcasting. I enunciate a measurable vision about what I want to have happen by a given time frame. I then work backwards to figure out what needs to be true for that vision to happen. What needs to be likely. And then, compare that back to what has happened in the past. Am I trending towards that vision? Am I regressing away from that vision?

Backcasting requires a vision. It assumes one knows where you’re going and why. But it also requires one to check their optimism with realism. Or check their pessimism with historical record. It can be extremely scary. Those being held to account will always want low targets. In so doing, they can claim outperformance. Those setting the target frequently want it to be less about continuing trends and/or luck. They want to cause an incremental delta on top of the existing trend.

Those are the differences. And they’re major.