The bug report was already in the request log

Post-launch, week two, the client forwards an email from their customer: checkout failed around 2pm. No screenshot, no details. Normally this is where you ask a stranger to please reproduce their worst experience of the day.

Instead you query the proxy log for errors around 2pm, request bodies included. There it is: the exact POST that failed, payload and all, timestamped 2:07. You boot a copy of production from last night's snapshot, replay the captured request with curl, and watch it fail on your terms. Fix, verify against the same request, deploy.

The client's customer never filed a proper bug report: and never needed to. The bug report was already in the log, body and all.