Is it really so hard to just ask people? No need to spy on them. Also I don't think spying on them will improve the product. Telemetry has limited value in the first place.
Having worked on a variety of user-facing products for several years, I can tell you that it is hard. It depends on your audience but typically many users don’t report problems. If they do, they don’t give you enough details to diagnose the issue. And why should they? They’re not knowledgeable in how your system works. So you have to reach out to them and ask for more information. Maybe they need to reproduce the problem to give it to you. If you’re unlucky, they won’t be able to reproduce it. Even if they were able to, you have now taken more of their time.
The ask could be as simple as “hey this went wrong, but you can click here to send us the error info and someone will get back to you” and many users will STILL not share the information. You might think that they must not care that much but often times you’d be wrong. Users can be surprisingly irrational at times.
The right time to ask if anyone will be using a specific portion of the app is when the feature is being proposed, not after it's been implemented and released. Replacing discovery up front with post-release telemetry is a sure way to lose money.
Conversely, if you're not making that mistake, then you don't really need telemetry to tell you what you already know.