Blog Business forecasting - tips from behavioral economics

Business forecasting is challenging for B2B companies, because B2B unit sales rarely support quantitative forecasting methodologies. Hence the importance, for such companies, of careful sales pipeline management and analysis.

This lack of large data sets doesn’t mean, however, that business forecasting should remain all art, no science. We have much to learn from behavioral economics and decision theory, from the sophisticated “quantitative debiasing of judgmental forecasts” to simpler, but equally efficient tricks.

As an introduction, I wanted to present 5 tricks validated by recent forecasting research and easy to implement in modern CRM software.

1. Explanation

Forecasts supplemented by a one-sentence rationale are more reliable, just because they are less mechanical. You should revise the rationale with each forecast update (ex. “moving the closing date from 31.03 to 15.05 because of recently surfaced compliance issues”).

2. Pros & Cons

For each forecast, list 3 reasons it should come true and 3 it shouldn’t. While the 3 “pros” are just a more detailed version of the “Explanation” trick, the 3 “cons” are useful overconfidence deflators.

3. Counter-valuation

The average of 2 (informed) judgmental forecasts (for the same event) is more reliable than any of its components. This is quite an important insight, as many B2B companies use “management overwrites” in forecasting. Research implies that averaging the forecasts of the sales rep and the sales manager will be more efficient than merely replacing the former with the latter.

4. High & Low

The average of a “best case” and “low case” is more reliable than a point forecast. Combining this trick with the “Explanation” trick results in an interesting variant of the “Pros & Cons” trick.

5. Multiple methods

One of the most robust and significant conclusions of forecasting research over the past 25 years is that combining multiple methods of forecasting is always beneficial. Indeed, the most sophisticated methods very often yield inferior results to combinations of 2-3 simple methods. Here again, the implication for business forecasting is important: using some form of weighted pipeline is fine, but you should add other forecasting methods (e.g. simulations, ideal pipeline) to your arsenal.

What do you think –  have you found other tricks useful?