Increasingly, political campaigns use digital ads to seek out potential voters, and companies like Meta offer advertisers detailed targeting options such as criteria based on user demographics, behaviors, and interests. An underexplored feature of digital advertising on social media platforms is the use of so-called ad delivery algorithms. These algorithms set prices via ad auctions and deliver ads to “relevant” audiences without the explicit knowledge or intention of advertisers, prompting the question of whether this process may constitute an algorithmic form of microtargeting.
To investigate the delivery and pricing of political ads, we collaborated with three Dutch political parties to place on their Facebook and Instagram accounts a total of 135 identical ads targeting nine different audiences before the country’s nationwide 2022 municipal elections. Ads ran with the same settings, at the same time, using the same daily budgets, texts, and images to ensure any differences in pricing and delivery are due to the advertiser and target audiences.
Our pre-registered hypotheses expect that political parties reach more people and pay lower prices when ads are targeted at “relevant” audiences. We find pricing and delivery differences between parties and audiences, but not always as expected. We find evidence that some parties are charged more than others, with one party paying 9.24% to 10.74% less to reach 1000 users. Furthermore, lower-educated citizens, women, and younger people (18-24 year olds) are more expensive for political parties to reach.
These findings have significant implications for political parties and democracy. The fact that parties are charged different prices for the same reach creates unfair competition and an uneven playing field. Our finding that certain groups are less likely to receive political ads means they are potentially more isolated from receiving election-related information on Meta platforms. Finally, our findings suggest that simply banning or severely limiting targeting capabilities, without addressing the potential for algorithmic microtargeting, would mean that ad delivery algorithms still deliver ads to specific groups without oversight or transparency.