Completion Is Not Capability

Completion rates survive because they are easy to collect, easy to report and hard to argue with. They tell us something real: people showed up and moved through the material.

They do not tell us the thing we actually want to know.

A more honest set of signals is usually available, even without new tooling:

• Can the person do the task unaided, once?
• Do they ask better questions afterwards?
• Does the error the training targeted appear less often?
• Does their manager notice a difference without being prompted?

None of these are as tidy as a percentage. All of them are closer to the point.

If a metric would look identical whether or not anyone learned anything, it is not a learning metric.

The practical move is small: keep completion for operations, and add one behavioural signal per programme. One is enough to change the conversation.