The Capture Rate

Last week I published Return on Corrections, RoCor: the value an organization keeps from its own correction stream.

Today, the first of its four rates.
A correction is born every time a human catches what a system got wrong. An override. A deferral. A dispute reason quietly renamed. And in AI-assisted workflows, corrections multiply: every output someone fixes before sending is one.

Most organizations fix and forget. The fix works once. The knowledge dies where it happened.

Capture Rate asks one question: of all the corrections your people made this month, how many exist anywhere on record?
Ask it Monday morning. You need no system, no budget, no project. Count two things for one week: how often your team overrides a system decision, and how often a case is deferred for later. That count is your baseline.

In my experience, most organizations capture less than 1 correction in 10. The other 9 they will pay to relearn.

Because the arithmetic is merciless:
What you do not capture, you cannot reuse.
What you cannot reuse, you buy again.
Every month. At Full Price.

Next in the series: the Reuse Rate.
The full metric, its science and its open questions: first comment.