A method for recommending content based on user profiles clustered by subscription data includes receiving usage data of users. The usage data for each user includes device usage data and subscription data associated with the one or more applications installed on the user device. The method further includes receiving a recommendation request from a first user device of a first user for new subscription data and determining first usage data of the first user. The method also includes identifying a second user having second usage data that has a threshold similarity to the first usage data amongst the users. The method includes determining the new subscription data as being at least some of the subscription data of the second usage data that is different than the subscription data of the first usage data and outputting the new subscription data to the user device of the first user.