We study a system where many identical users of a service share a common resource. Each user is sensitive to congestion at the resource, but also experiences a positive network effect. We consider a model where both effects depend on the usage of individuals in the system, as well as potentially the number of users in the system. We consider two benchmark scales for the service: the subscriber base most preferred by an individual user, and the subscriber base most preferred by the service manager. The latter can be significantly larger than the former; we discuss the consequences for management and operation of congestible services.