We should store synthetic testing metrics longer than we do today. The crux metrics we collect we store for 2 years, but for our own metrics we save them only for 33 days. That should at least be one quarter (so we can compare the beginning of the quarter with the end) or even better one full year.
We also have a minimum retention of 10 minutes but we do not run tests that often, I think we should change that to 1 hour and the we can setup specific rules if we need to run tests more often.