Breakthrough vs control — two different jobs, and why most teams only do one

Control is the world of monitoring charts: watch the numbers, know the size of the normal day-to-day bounce, react only when something moves beyond it. Success, for control, is nothing changing.

Here's the uncomfortable part: most of your waste lives comfortably inside the normal bounce. It's produced reliably, every day, by the process working exactly as designed. No amount of reacting to individual bad days touches it — because the bad days aren't the problem, the design is.

Breakthrough attacks that built-in level — and the test of a real breakthrough is statistical, not anecdotal. Has the average genuinely shifted by more than the noise? The standard check compares before and after:

In plain words: take the gap between the old average and the new one, and ask whether it's big compared to the bounce you'd expect from chance alone. A gap that's large relative to the noise is a real shift. A gap within the noise is a good month wearing a medal.

• All control, no breakthrough: immaculate dashboards faithfully guarding a mediocre level.
• All breakthrough, no control: the workshop delivers, then the gains quietly rot away, because nobody set up monitoring at the new level. An improvement that decays was a demonstration, not a change.
• Moving the goalposts: the "improvement" appears because someone changed how the metric is counted. Keep definitions identical across the before/after line, or the comparison means nothing.

Plan the two jobs as a pair, explicitly. Breakthrough to move the level: a project, a hypothesis, honest before-and-after evidence. Control to bank it: new limits, a response plan, a named owner. Half the pair delivers roughly none of the value.