Every job is profitable until someone adds up the hours

Estimating is the core commercial skill in a job shop, and in most shops it is calibrated by recollection. The jobs that are remembered are the disasters; the ones that quietly ran twenty percent over are not remembered at all.

Booking time is the obstacle

Operators resist it, reasonably: it interrupts work and it feels like surveillance. Booking at operation level rather than in fine increments, and using it visibly to fix estimates rather than to assess people, is what makes it stick.

The comparison is the whole value

One job over its estimate is noise. Twenty jobs of a type consistently over is an estimating rule that needs changing, and that is a finding no amount of experience produces on its own.

What it protects

Usually a specific category — small batches with long setups, or a material that machines slower than assumed. Shops frequently find they are winning that work precisely because they are the cheapest, and cheapest for a reason they had not identified.