Is It Beneficial to Recommend Differently Priced Products? Experimental Evidence from an Online Product Recommendation System

Abstract

Online recommendation systems recommend products with widely different prices than that of their focal products. While conventional wisdom suggests that consumers may prefer lower priced recommendations, prior literature also indicates that consumers may not accept such products if their prices fall outside the range of their reference prices. We empirically examine this question – how does recommending differently priced product affect their demand – with a field experiment on a US based fashion retailer's website. We find that recommending differently priced products decreases their purchase probability by 12.5 percent. We estimate several exacting specifications to show that our results are due to the differences in prices and not characteristics between the focal and recommended products. Based on our estimate, we simulate the demand for recommended products by replacing the lowest order differently priced recommendations for focal products with the similarly priced products. Such replacement results in 23 percent increase in the purchase probability of recommended products, which translates into a 2 percent increase in the total sales of recommended products. Overall, our study highlights that the relative price of recommended products could significantly influence their demand and therefore, it should be considered as an additional factor in design of recommendation algorithm.