Is that what you were referring to? The phrase "we have to collect customer data" implies a different thing entirely, so I misunderstood.
My objection to Microsoft's methods in this regard isn't the data that customers voluntarily and knowingly store on Microsoft servers, it's the collection of data about customers, their machines, and the use of their machines that happens behind the scenes.
We collect telemetry about user actions and the successes of our service. All the data we collect is about how the service runs and we only look at it via aggregation. Internally we have training every year about what you can and can't collect and under which scenarios. It gets stricter every year.