> Then I would completely break the feature out into its own app with separate branding. Hopefully I could get 10-20% of the active users of that feature to move to the new app and then grow it from there.

As we all know this strategy worked great for Foursquare.