Is it really that cost prohibitive? Actual numbers would be useful to make sense of this discussion.

The question comes down to value. If you’re never actually using the data, fine. But once you need it, and you don’t have it, might the cost sometimes be greater than it would have been to preserve it?

The choice of how much telemetry to keep seems similar to how much you want to spend on an insurance policy. The same kinds of risk evaluation apply.