I asked for actual concrete evidence, not "can be used". Is there an example of a program that was crap, implemented telemetry and then got better afterwards?

I asked for actual concrete evidence, not "can be used". Is there an example of a program that was crap, implemented telemetry and then got better afterwards? (and of course controlled for factors that might have improved the program anyway) I mean since telemetry advocates are so into how useful data is, surely they must have data on whether telemetry itself works?