โ† All tools

Signal vs noise

Manage a team through eight quarters of noisy results. You'll be quietly trained to chase noise โ€” then shown, at the end, how little of it was ever yours to manage.

  1. Observe
  2. Choose
  3. Reveal
  4. Repeat

Open a conversation with anyone whose number you think reflects something real โ€” or leave the rest alone.

Why this fools you

A quarter of anyone's output is one noisy sample from a person-in-system process whose baseline the system sets โ€” scope, dependencies, luck, drag. Deming's claim was that the system dwarfs the individual as a source of variance. React to each swing as a special cause โ€” a real, assignable signal โ€” when it's just this month's draw, and you're tampering: adding variance, not removing it.

And regression to the mean makes the tampering feel vindicated. Praise your best quarter and it falls back โ€” "praise breeds complacency". Warn your worst and it bounces โ€” "tough love works". Both are pure regression; you learn to manage exactly wrong. This is the flight-instructor illusion, and here it happens to you.

The routine-variation band is an oracle โ€” you only get it in a simulation; your real team doesn't come with one. What transfers isn't the band, it's the question every swing deserves: spike, or shift? A single point outside the band can be luck; only a sustained walk out of it is a signal worth a conversation.