Naive analytics: The strategic advantage of algorithmic heuristics

We study interactions with uncertainty about demand sensitivity that is estimated by analytics algorithms. In our solution concept (1) firms choose seemingly optimal strategies based on estimates from possibly biased analytics algorithms, and (2) the levels of biases form best replies to one another. In equilibrium the firms' algorithms overestimate advertising effectiveness, as observed empirically, which causes advertisers to overspend. In price competitions firms also underestimate price elasticities and set prices too high. In games with strategic complements (substitutes), profits induced by such “naive analytics” equilibria Pareto dominate (are dominated by) those induced by the Nash equilibrium.