My package, poppr, has been on CRAN for over six years now and it has received more than 400 citations and just north of 70,000 downloads from the RStudio cran mirror. I can think of three reasons why this package has been successful:

1. Our lab gave several workshops using our package over the years
2. We have written extensive documentation with both web site and package documentation.
3. I actively maintain and answer the vast majority of questions that appear on the poppr google group.

I think this last point is crucial. The poppr package has my email address attached to it in the Maintainer field, which is the person you should contact if something goes wrong, but if anyone sends me an email about poppr, I will always¹ send them to the poppr forum (or the GitHub issue tracker if it is a bug). Even if the user reports a really embarassing bug, I will do my best to make sure it gets fixed and sees the light of day as soon as possible. I do this because the best way to build trust with your users is to be brutally honest in the most transparent way possible; the best way I know to do this is through an official forum.

For users, workshops and documentation are nice introductions, but a forum is really one of the few places where you can actually engage with the developers and the community to get into nitty-gritty details of an analysis, get help with some unclear documentation, or report some strange behavior that may or may not be a bug. From a developer perspective, a forum is an invaluable tool that allows you to see how your users work and shows you a world of analysis beyond what’s inside your head or in your immediate workgroup. I’ve had people ask about clever workarounds for violated assumptions, report bugs, and request features that eventually became part of poppr. The best part of all of this is that it is all public, which means that it’s possible for users to find answers to their questions before they even have to ask it.