Targeted Advertising as an Implicit Recommendation and Personal Data Opt-Out

Abstract

Advances in data collection and algorithms help advertisers to better target individual consumers by predicting each consumer's preferences. We first show that when consumers have uncertainties about their preferences, an ad targeted to a consumer carries an implicit message: the algorithm predicts that the product fits her preferences. This implicit recommendation influences the consumer's purchase decision but also introduces misaligned incentives. As the accuracy improves, consumer inference from targeted ads becomes stronger, but so does the advertiser's incentive to exploit it to affect the consumer's decision. Under exogenous price, when individual-level prediction becomes more accurate, the advertiser adopts a less targeted advertising strategy due to its enhanced incentive to exploit a stronger recommendation effect. Even if the firm's prediction is perfectly accurate, consumers still receive ads for "bad products" and make incorrect purchase decisions. Despite these negative consequences, the consumer surplus can remain positive because the firm can better identify consumers with a good fit for the product. In contrast, under endogenous price, a better prediction allows the advertiser to raise its price instead of exploiting its recommendation effect. Thus, it leads to more targeted advertising and lower consumer welfare, which may incentivize consumers to opt-out of data collection.

Keywords: AI, algorithms, targeted advertising, recommendation, persuasion, pricing, privacy choice, personal data opt-out

JEL Classification: M37, M38, C72, D82, L00