I would try to define key points that are used as a tester.
Time by Ticket/Story, Prevented Regressions, overall coverage .... you have to pick the code qualities / product qualities of the product and think about what you can connect with your actual work.
Sorry that I can't say more about this because I think as always the devil is in the details and you would have to track "correlations to causation" what went wrong and what could we do to prevent it. and than you need to categorize it and try to create a graph and find out if your correlations are actually connected to the causalities.