Peer Effects from Friends and Strangers : Evidence from Random Matchmaking in an Online Game

We empirically study how to seed products in a network in the presence of peer effects from both friends and strangers. The product adoption decisions of friends and strangers can both strongly influence consumers' purchase behaviour, but their importance relative to each other is unknown. We first collect novel data on microtransactions in an online gaming app to quantify the lift in purchase probability from observing friends' and strangers' product adoption choices. To isolate exogenous exposure to peers' product adoptions, we leverage the app's quasi-random matchmaking system. Observing a friend's product adoption has twice the marginal effect on purchase probability as observing a stranger's product adoption, but the cumulative effect from observing strangers' product adoptions is much larger over time. We then calibrate a diffusion model using the in-app social network of 9.4 million players. We find that seeding products to active players who interact with more strangers generates almost four times the number of net new adoptions compared to seeding to players who are more central in the social network