Several different guidelines, laws are mentioning requirements about deviation management, but mostly about the essence of classification, identifying the true root cause, etc. But it's not described what can be a good foundation of all of these: event detection and initial investigation.
Whatever type of deviation we discover, it's our own responsibility to collect as much information about the phenomenon, as possible. The classic 5W2H tool can help us out to sort out what we need:
What (exactly happened)?
Who (has been there? who performed some activity in the vicinity? who might have more information about the event)?
When (did the event happen)?
Where (did the event happen)?
Why (may have it happened - in other words, what may be the direct cause of the event)?
How (may have it happened)?
How many/How much? - any numerical data collected during the event (think creatively: how many people were there? what was the temperature in the vessel/in the area? how much of a certain material was in place?)
+1: is there any GMP product potentially impacted? if yes, what are their batch number or any other ID?
Don't forget, you need to be fast! The environmental conditions can change quickly, and some evidence may disappear forever.
Don't waste your time with talking it through with multiple people in meetings, etc., just register the event in your deviation management system with all data collected (the industry practice is usually 1 day from discovery), and then you can think through who may be the best person/people to perform the systematic investigation.
I believe the same challenge is there also related to the OOS-investigations.
If you are to answer the usual (and following MHRA and FDA OOS guidances, mandatory) questions, like 'Did the analyst use the right volumetric tools to carry out the test?', you won't have any chance to get that answer e.g., 3 days after getting the OOS - everything has been washed up, the solutions have been poured out to waste, the analyst did 10-50 tests since the OOS, they will remember nothing (and that's not their fault!).
So to sum it up, don't waste your time on bureaucratics, don't pass the registration of the event to someone else, do it yourself, if you've discovered it!
Simply because at this stage, you yourself know the most about the phenomenon, don't let the collected information go to waste.