The idea is to find the different kinds of users so you can develop personas.
To do that focus on the value your product provides to different people. For example if one persona is a very busy person living in a non-English-speaking country with relatively poor/flaky 3G connectivity and not much time to use any app (much less yours), that can help you tune your app to be appropriate for her use. If your app is not localised and depends on an always-on Internet connection, your app is currently not meeting her needs very well.
The decision you need to make (and the reason you generally do this work before design and development) is to figure out if it's worth it for you to target that kind of user; whether the value to you/your client is enough to justify the effort to support her.
If you can't eliminate any kinds of users as explicitly outside the scope of your app, you're going to have many, many personas and it's very quickly going to be unwieldy. The trick, then, is to pick the top 4 or 5 based on your own ability to support them to a high degree.
If that means eliminating, for example, the 75-year-old male user who can benefit greatly by using your app, but who cost you a disproportionately-large amount to support as a customer (due to their specialist needs) you can do so.
Many people have difficulty establishing their personas, but it seems to me that it usually happens because people instinctively hate the idea of limiting their potential market. The purpose of personas is not to restrict your market to a select few individuals or people like them so much as to make sure you're focusing your attention on the differing needs of users like them who you've identified as a market you can support profitably. Removing that persona as a target does not mean restricting access to those kinds of users; it simply means prioritising the user group where you see the most valuable opportunities to address a need.
