I've seen people tune their apps for these early adopters, talking about how they've got their "viral coefficients" at the correct rate (via product improvements, design, marketing efforts, etc).

Then, by induction, they think the same metrics will hold true once they launch to the general public.

Then they launch and are surprised to find that the numbers don't apply to later categories of users.

I'd be curious if there's a model to solve for that problem.