Regression to the Mean
Did the fix work, or was the bad month always going to end?
The idea
Extreme results are usually part luck, so the next measurement tends to sit closer to the average whatever you do.
In the real world
The worst-performing month improves after an intervention that did nothing.
Going deeper
Extreme results are part signal and part luck, and luck does not persist. So the measurement after an extreme one tends to sit closer to average regardless of any intervention.
This is why treatments applied to outliers appear to work. Coaching the worst-performing team, changing the process after the worst month, promoting after the best quarter — all are followed by movement toward the average that would have happened anyway. Selecting on an extreme guarantees an apparent effect, which is why attributing anything requires a comparison group that was equally extreme and left alone.
Where it stops applying
Regression does not mean interventions never work; it means the naive before-and-after comparison cannot tell you whether this one did.
Why it matters
It is the main reason interventions applied to outliers appear to work.
Try this today
Before crediting a fix applied after a bad month, ask what an average month would have looked like anyway.
Test yourself
A team is coached after its worst quarter and improves the next one. Management concludes the coaching worked. What else explains it?
Show the answer
The worst quarter was partly bad luck, and luck does not persist, so the next quarter would likely have been better regardless. Selecting on an extreme guarantees an apparent improvement, which is why a comparison group is needed to attribute anything.
Learn this in the feed Answering from memory, then again days later, is what makes it stick.