I only came across the matrix recently. Codex recommends using their decision tree. The thing which confuses me about the likelihood / severity thing is should you do it before prerequisite controls are in place, before
CCP controls are in place, after all controls or all three? I suppose if you do a "before" and "after", it might tell you whether you have controlled the hazard, however the severity will never be reduced, only the likelihood.
Personally I use the decision tree but discuss it as a team, otherwise you can end up with a ridiculous number of
CCPs. What I always think if the decision tree comes out with a 'yes' for a
CCP is "is the action which controls this specific, well located and measurable? Would it be controlled better as a prerequisite?" A hot topic for example is cleaning. Some people would identify this as a
CCP. I think it's a prerequisite because it happens in lots of places in a plant, it can easily be bypassed whatever controls are put in place, it's difficult to monitor (ok there is ATP swabbing but it's not very accurate and depends greatly where you swab, there's chemical concentration and temperature testing but that's a waste of time if the method is wrong). However, a decision tree will call that a
CCP if you ask the questions in a certain way.
