Do Referral Programs Drive Loyalty?

Using unique data from a leading Chinese content platform with more than 300,000 users, we propose a structural approach to evaluate the effect of the structure of a referral network on users’ renewal decisions. Referral networks provide essential identification sources, which enable us to embed the expectation of network peers’ behavior into the utility function as an important component to capture the decision variations. We find that these networks play an essential role in users’ renewal decisions, which are significantly and positively associated with the renewal decisions of both referrers and referrals. Our counterfactual analysis has important implications for the referral policies of digital platforms. First, we find that the referral-targeted discount discrimination policy is more effective than the uniform discount policy. More optimistic expectations for referrals’ decisions due to the price discount generate a snowball effect on referral networks, which in turn increases renewal rates. Compared to a uniform discount policy, a more referral-targeted discount policy would significantly increase renewal rates while reducing overall revenue loss. Second, our results highlight the importance of the structure of a referral network. With the same beta index, a high-centrality network implies a reduction in the chain hierarchy, which is detrimental to customer retention. We suggest that an efficient referral network should be highly connected with a lower degree of closeness-based centrality.