If many regular TV viewer were switching to over-the-top services it would show up in the numbers. Instead we have data points like the Wired article
that showed about 300k Netflix users consumed traffic that represented 20% of total US peak Internet traffic on a nightly basis. If that had ballooned out to 3m streams on a nightly basis, the numbers would be far more staggering and reported regularly.
I think the story here is far less interesting: people who don't watch a lot of TV canceled their pay TV subscription.