I don't really think this makes sense - they pay a relatively large engineering team, and they spend an inordinate amount on purchasing content to get engaged v

I don't really think this makes sense - they pay a relatively large engineering team, and they spend an inordinate amount on purchasing content to get engaged viewers. The cost of servicing the viewers is probably fairly trivial given the peering agreements they currently have in place and their distributed content infrastructure. I think the real problem is that if you're very measurement based - you make poor product decisions. You might be able to optimize all sorts of metrics, but if you can't measure it (or measuring it is hard/noisy/laggy), you tend to lose sight of it. In this case - Netflix isn't adequately measuring consumer satisfaction, and they haven't ever really been able to measure show quality without long term viewership numbers. At best, they correlate show quality with viewership, but that's a tricky game, and it's often a very laggy measurement - viewership at release is a very poor gauge. Netflix became the behemoth they are today by focusing on the long tail of shows - shows that had consistently moderate viewer counts over a long period of time - rather than focusing on boom and bust blockbusters. Personally - I think they're fucking up. I also cancelled this year after more than a decade.