Diversity in Advertising in Times of Racial Unrest

Abstract

Firms increasingly aim to present diversity across age, gender, and/or race in their promotional material. Recent social movements such as BlackLivesMatter (BLM) or MeToo have further raised awareness towards social inequality. This paper examines racial diversity in U.S. digital advertising, analyzing both the diversity of ads distributed to consumers (supply) and consumer reactions to diverse ads (demand). We also investigate the change in advertising diversity (supply and demand) surrounding George Floyd’s murder and the subsequent BLM protests. Using a longitudinal dataset of tens of thousands of display ads, generating over 250 billion impressions, we employ robust, advanced deep learning methods to automatically detect demographic diversity in these ads. From January 2019 to July 2021, we observe an overall increase in minority models in digital ads, particularly Black models, along with higher click-through rates for ads featuring Black models. The murder of George Floyd led to a temporary surge in Black model inclusion and more positive consumer reactions. However, both effects subsided within a few months. Overall, our findings indicate that diverse advertising can create a win-win opportunity, promoting minority representation in public media while benefiting advertisers financially, and that social movements like BLM can influence the supply and demand of advertising.

Keywords: diversity, online advertising, social inequality, racism, machine learning