A social feature spreads through friends — how do you A/B it?

Experiment designHard

Problem. You're testing a feature that makes it easier for friends to join each other's experiences. Design the experiment — in particular, what do you randomize, and why is a normal user-level A/B wrong here?

Before you reveal: say your answer out loud, as if you were in the real interview — get your reasoning across clearly first. There is no single correct answer: reading what the interviewer is really after and defending your own thinking is what makes an answer strong.