If you give your employees privacy they will experiment and adjust the way they work to their context and experience without trying to guess your opinion first. If you can't leave them unsurveilled without fear of mistake or fraud then you don't have a surveillance coverage problem, you have a deep hiring and management problem.
If you give your users privacy they will trust you with their secrets because they will know you don't remember them. If your business model needs you to spy on them because you don't dare asking then you aren't serving them, and they will know and reciprocate your loyalty.
The acquisition of information has costs: more technical resources, more processes complexity, more reputational risks, more exposure to bad faith actors, more tension in your relationship with users and employees. It may be a cost worth paying, but it has to be budgeted, and it should be paid in the open.
Privacy is thus a wu wei feature: you make your system better simply by not doing what you don't need to do. And ethics aside, by this point it should be clear that the myth of "big data" on its own leading to improved outcomes is that, a myth. A large data lake is no more guarantee of intelligence than a large library, with the difference that data lakes now have the reputational glamour that libraries used to. Data is easy, understanding is hard - thus the perpetual temptation of adding just one more data point.