Reversion to the Mean (Business Performance)

Extreme business results may become less extreme when temporary conditions or noise fade, and profitability may also change through competition and adjustment. Statistical regression to the mean is distinct from those causal mechanisms and does not guarantee a return to an industry average.

Extreme observed performance can combine durable capability with temporary conditions and measurement noise. When a company is selected because of an unusually good result, the temporary component may be less favorable next time. This statistical selection effect does not require a competitor or manager to change anything.

Business performance can also change through causal processes such as entry, imitation, restructuring, or shifts in demand. Keep those processes separate from statistical regression and estimate persistence using suitable comparisons. A historical average is not an automatic destination: the business, industry, and measurement can change, and poor performers can deteriorate or disappear.

When to use it

When a company is selected for extreme performance, a forecast extrapolates a peak or trough, or an intervention is credited for a change from an unusual baseline.

How it can help

Encourages separating persistent capability from temporary conditions and selection effects when evaluating results or forecasting performance.