The gold standard for reliability studies involves collecting in situ failures though. You can infer a lot from inspections on non-failed components, but it's n

The gold standard for reliability studies involves collecting in situ failures though. You can infer a lot from inspections on non-failed components, but it's not as useful as having actual failure data. When I'm referencing their in-house data, I'm referring to failures (whether in situ or as a bench experiment)