Compare the two scenarios:
Scenario A: “We received a piece of user feedback to change this design to avoid errors; their suggestion was well received by other users. Let’s collect some telemetry from that component to see is these issues are representative of the larger population; based on that, we’ll know whether it warrants a re-design of that component.”
Scenario B: “Telemetry says users never use this feature; we can remove it.”
In Scenario B, telemetry prompted a decision; in Scenario A, telemetry helped understand a real specific problem. Telemetry should be used to clarify an existing insight rather than make discoveries on its own. Metrics should not be collected lightly; they should be collected with intention (and, of course, prior informed consent).