You may be right, but people leaving the site isn't really the key metric here. The key metrics are ad and subscriber revenue: in this case particularly ad revenue.

If the reason they're leaving makes it's more likely they'll come back again in future this may lead to an increase in ad revenue due to more sessions even though average session duration falls.

Obviously I don't know, and unfortunately I've learned not to trust that higher-ups in certain organisations have done their homework on revenue-affecting decisions, but you'd at least hope that they had some data to back up the thinking.