Originally published at https://sazm.in/articles/fix-production-bug-in-48-hours-async
Most production bugs don't stay open because they're hard. They stay open because the fix keeps waiting on people: a call to explain the problem, another to grant access, a third to confirm it works. Every handoff adds a day.
After 20+ years in production (since Oct 2006), this is the fastest pattern I know. It's fully written and fully async:
- Write a bug report a developer can act on. Cover what happened, what you expected, steps to reproduce, when it started, and evidence (screenshots, console text, failed order IDs). Those steps later become the acceptance test.
- Grant scoped, revocable access once. Use a staging copy, separate accounts for the developer, test-mode payment keys, and a fresh backup before anyone starts.
- Agree a fixed price and a definition of done. Get a root-cause hypothesis, a fixed price, a delivery window and an acceptance test in writing before any code is touched.
- Accept written proof, not a "done" message. Ask for the root cause, the list of changes, before/after evidence and a rollback note.
- Stop the bug from coming back. Run updates on staging first and keep daily off-site backups.
The full guide, with a copy-paste bug report template and an FAQ, is here:
👉 https://sazm.in/articles/fix-production-bug-in-48-hours-async
If you have a bug you'd rather hand off, I do fixed-price fixes within 48 hours, entirely by email: https://sazm.in/bug-fix-services · past work: https://sazm.in/projects
What's the slowest bug fix you've been stuck waiting on, and what held it up?