United in Diversity, Divided by Algorithms: Ad Delivery Algorithms in the 2024 European Parliament Elections

Social media platforms like Meta serve as critical infrastructure for political advertising, with their automated systems determining the distribution of campaign messages to millions of voters. Although studies have examined the use and effects of social media advertising, empirical research on ad delivery algorithms remains limited. Understanding these algorithms is essential, as they make decisions about message distribution and pricing that can introduce systematic biases, potentially threatening the fairness of democratic processes in the digital public sphere. While traditional media and campaign regulations have long recognized the importance of maintaining a level playing field to ensure fair electoral competition, comparable frameworks for social media platforms and their ad delivery algorithms are rare or non-existent. Using a pre-registered large-scale algorithm audit study involving the collaboration of 30 political parties across 8 European countries, this paper investigates how Facebook's ad delivery algorithms influenced the pricing and distribution of political advertisements during the 2024 European Parliament elections. We ran identical, non-partisan "get-out-the-vote" advertisements from official party accounts to test for pricing and delivery disparities. Our research design varied only the advertising party and target audience of the ads that ran at the same time, with the same settings and budgets. Our findings reveal striking disparities in ad costs and reach between countries and parties. We observed price differences of up to 332% between countries and up to 27% between parties within the same country, leading to substantially less reach for the same price. Our analysis of account-level factors suggests higher costs for parties with substantial ad spending history, hinting at a potential 'success tax,' while party-level characteristics tend to show that a party's voter base are cheaper to reach. Market-level conditions, particularly the size of the target audience, emerged as significant predictors of ad costs.The uncovered algorithmic disparities highlight previously unknown opaque forms of inequality in digital political campaigning that may advantage some political actors while disadvantaging others. The findings inform ongoing debates about the need for greater transparency and accountability in digital political advertising and underscore the urgent need to examine how algorithmic systems shape democratic discourse in the digital age.