Correlation vs Causation

Correlation vs causation is the reminder that two metrics moving together does not prove one causes the other; the link may be coincidental, reversed, or the effect of a third factor.

This distinction is the most important guard against wrong data-driven decisions. Three scenarios let correlation masquerade as causation: a hidden factor (both metrics are effects of a third cause), reverse causation (the relationship runs the opposite way to what you think), and pure coincidence (especially when you mine dozens of metrics at once, some patterns co-move by accident). The only definitive way to prove causation is a controlled experiment: make a change, hold everything else fixed, and see whether the target metric moves. In observational data — which web analytics entirely is — causation is never definitively proven, only made more or less likely. The expensive mistake is leaping from a correlation to a big action without ever testing the causation.