Competition Through Recommendations

This paper examines how two-sided platforms develop their recommender systems to be precise about value-for-money. On each platform, more precise recommendations generate ranking and screening effects: they steer demand toward high value-for-money products, intensifying price competition among firms which drives out lower-quality firms. Thus, more precise recommendations benefit consumers but reduce platform’s per-transaction revenue. A monopolist platform still prefers precise recommendations, as this expands demand. Competing platforms choose even more precise recommendations. However, when consumers search across platforms or recommender systems are overly complex, recommendations become less precise. This shows that market power is only one potential explanation for 'ensh*ttification'.