The limits of '$ per review' as a code-review metric

A developer who runs Greptile alongside an internal code-review system calls Greptile impressive but says their own system catches more bugs and issues upfront, with greater correctness. They dispute Greptile's '$ per review' metric, arguing that it misses the extra PRs created as Greptile reveals new bugs incrementally. They consider '$ per PR' a better, though still imperfect, measure.