During OQ, a room that recovers in 22 minutes against a 15-minute criterion and a tester who skipped step 7 are two different deviations.
The first is a test deviation. The system did not meet its acceptance criterion. It gets raised, the root cause gets investigated, and the impact is assessed on other tests and on the system itself. Then it gets fixed, and the test is repeated, often with the tests that depend on it. Re-running until it passes, with no cause found, is the pattern an inspector looks for.
The second is a protocol deviation. The result may be fine, but the protocol was not executed as written: a skipped step, an instrument that was not the one listed, a prerequisite not met, a sequence changed on the day. The question is whether that affects the validity of the result. Sometimes it can be justified and accepted. Sometimes the test is repeated. If the protocol itself was wrong, it is amended through change control and approved before anyone executes it again.
Both need the same things on paper: who raised it, who assessed impact, and QA approval before the report closes.
Keep one trap in view. Changing an acceptance criterion after a failure looks like a protocol fix, and it is the first thing an auditor questions.
Commissioning and qualification sit in our EPCM work.



