A/B testing is really hard, even when there is no social interaction internal to your application (like a game). The easiest thing to test is marketing for new users. But even that can be tricky, where one group of people might click fewer ads but are higher quality. This is especially troublesome when you can't tell how many friends they invite through word of mouth. And then if you want to implement other features or fix bugs over the duration of the test, that also poisons the results.
And then the number of incoming users you need is massive.
To do a good A/B test that doesn't mislead you is extremely difficult. Almost to the point that for most companies, I don't think it is worth doing.
> So... I guess my final recommendation is that you should hire some data scientists that like doing experiments.
Like this guy says, you basically need a data science team to do it effectively. If you've got a small team (like I work with), don't waste your time with it. You're better off just adding new features and fixing bugs.