Problem is - it probably works. It gets a percentage of people to stay longer. You can argue that those full screen "before you leave" popups are annoying, and

Problem is - it probably works. It gets a percentage of people to stay longer. You can argue that those full screen "before you leave" popups are annoying, and who in their right mind will abuse users like that? Then they'll show you the results of A/B testing, prooving 15% more users subscribed thanks to that banner. We're screwing ourselves over statistics...