Technically, you're right, however each domain his its own notion of what "adequate coverage" means. Some projects use software code coverage. I deal with integrated circuit observability/controllability, and stocastic functional testing.

Until something, outside of the scope of unittest, becomes available (or affordable) having a count of the number of test points for each test gives me a good idea of what testing has been done and what needs to be expanded. Practically, having a count of the number of tests is a first-order useful measure for many of the projects I deal with.