From a platform perspective, a big metric for me is: What percentage of tests fail for reasons that aren't "developer's fault"?
I'd also be interested in test execution time, especially how early in the pipeline we can detect failures. (e.g. why couldn't this have been found before it got to CI?).
There are a ton more metrics I'd be interested in as a development team who owns the test (coverage, pass rate, etc), but that seems outside of the scope of what I'd expect platform to monitor.