On the other hand, maybe it was a mistake for the industry to start leaning so hard on telemetry. While it can provide useful insights in some cases, telemetry is best positioned as secondary or even tertiary in the decision-making process, because it doesn’t tell anywhere close the whole story and requires a great deal of error-prone interpretation to act upon.

To be frank, I find that use of analytics as the primary determiner to be incredibly intellectually lazy. They’re being used as a stand-in for deep, holistic thought in product design, user research, 1:1 interviews with users, etc when they shouldn’t be, and it’s making software as a whole much crappier than it needs to be.