Noah Kagan: Explaining the Basic Viral Model

Virality is only a third of the equation. The other two parts are the saturation size (usually a function of the market you're in) and the rate at which your userbase decays (do users visit once, invite ten friends, and then never come back, or do they come back every day?).

Without factoring at least these two things into your model the "viral coefficient" is mostly useless.

Use A/B split tests to increase X and Y - even a small increase in either can make a huge difference.

something else i've learned is multivariate testing is very effective. google optimizer can do this for free. services.google.com/websiteoptimizer/