> Maybe if I'd had data showing the actual time lost waiting?

This is exactly what works.

Adding analytics that reports duration of test runs to a central server makes it easy. Some developers panic at the thought of this because it feels like spying, but the data is immensely helpful.