Most people think of non-paying users (NPUs) in freemium apps as nothing but an unfortunate, avoidable expense; they eat up bandwidth and server space and, depending on the acquisition strategy, cost money to bring into the app. And while these points are true, denigrating NPUs as unfortunate is wrong-headed: no freemium app can survive without NPUs. NPUs contribute many things to a freemium app -- behavioral data, community vitality, and viral dissemination points. So while NPUs may be an expense, they're not unavoidable; they're more like a cost of doing business. Every freemium app needs non-paying users.
The freemium model is conceptually predicated on the idea that a free app, through wider adoption, will deliver a richer behavioral dataset than a paid app; that dataset can be utilized to extract more revenue from paying users than would have been extracted under the one-time payment model. NPUs are a crucial component of this concept; if data is the key to extracting revenues within the freemium model, then data from NPUs is valuable because it provides a contextual juxtaposition to the behavior of paying users. Some developers take the stance that NPUs should be punished after enough time with insurmountable payment walls or invite gates. This attitude ignores the value that NPUs provide to apps, which can be divided into three categories:
1) Behavioral and demographic data. Understanding the profile of users that don't pay is as important as understanding the profiles of those that do, because non-payers serve as the concrete, immutable lower bound on a continuous range of payment values. Instead of classifying users into binary groups (paying and non-paying), users should be placed on an LCV curve representing the revenue values they might eventually contribute to the app (with users that will never pay stationed at the 0 value). Under this perspective, having a constant against which mechanics can be compared is incredibly valuable.
2) Cross-promotion. A paying user in one app doesn't necessarily pay indiscriminately in every app she downloads. A user might engage with an app every day for months or years without ever paying; this behavior doesn't mean the user is cheap or disinclined to pay ever, it just means the app hasn't powerfully resonated with the user. Some apps simply aren't great monetization matches with certain users, even if those users enjoy the app and use it frequently. Removing non-paying users from an ecosystem is a mistake; instead, these users should be relocated to new apps within the developer's portfolio until they're matched with something they find appealing enough to monetize with.
3) Nightclub effect. Imagine a nightclub in which everyone is having a great time, and a group of people in the back is enjoying bottle service. The nightclub manager knows that some percentage of the people on the dance floor haven't spent any money on drinks, but he won't kick them out of the nightclub. Why? Because the people spending the most money -- the ones who bought bottle service -- bought a premium product precisely because it exhibits a clear value distinction from what is available to everyone. The standard product in a freemium app is the basic, free functionality; everyone that wants to feel superior to the standard players will purchase in-app virtual goods to enjoy a short-term sense of advantage. Furthermore, no one goes to an empty nightclub; not only do non-paying freemium app users serve as good cannon fodder for the paying users, but they also (in multiplayer games and games with strong communities) create artifacts of human participation that relay to the world that the app is being used.
Forcefully ejecting non-paying users from a freemium app negates the benefits provided by the freemium model. NPUs serve a purpose, even if that purpose doesn't directly relate to increased revenues.