Rigor Is a Budget: Match Your QA Effort to What Failure Actually Costs

A software engineer who builds both personal websites and healthcare systems argues that process rigor should be calibrated to the real-world cost of failure, not applied uniformly. On solo marketing projects he skips pull requests entirely, while a Medicare reimbursement engine underwent 57 rounds of QA before release. A denial-assessment engine was held behind a quality gate requiring 92% classification accuracy, with action accuracy climbing from 36.7% to 76.7% across four evaluation runs before shipping. One late QA round caught a critical bug where a patient-facing QR code pointed to a corporate login wall instead of the patient app — a flaw that passed all local tests. His core principle: price your process by what breaks when the system fails, not by convention or habit.