People seem to frequently forget that your measurement is always a proxy. A proxy for what? That's for you to decide. But you can't ever measure anything directly.
It's sadly common even in data science but the biggest offenders I see are business people. You can't use metrics without understanding what they're for otherwise you'll just enshitify your product. Be that number of clicks, time on page/site, number of academic citations, whatever. They're meaningless at face value. A good example is time on site. That's great for social media like YouTube where it generally correlates strongly to "using the website for what the website does" but even then it can be "scrolled for 20 minutes looking for something to watch, gave up. But on a news site, no one fucking cares about that. Your job is to provide information, so time on {page, site} might actually be a negative because it doesn't necessarily mean people are reading that whole time but they might be struggling to find the information they are looking for. A confusing UX also optimizes time on site.
https://en.wikipedia.org/wiki/Goodhart%27s_law