Year: 2025
Type: conference-paper
Abstract: Health authorities who wish to design a health-promoting advertising campaign would like to create high quality, trustworthy ads that cause behavior change. Here we compare human copywriters' ads to those created by other methods with different levels of expertise and trustworthiness, including Large Language Models, non-specialists, and students of a marketing class (with some expertise). To do so, we ran an online search ads campaign for 10 health topics with 142 ads from five sources and measured ad performance through the Click-through rate (CTR). Our results show that the copywriter ads' CTR was no better than the CTR of ads created by a Large Language Model, by marketing students, or by nonspecialists. The CTR of the best marketing student's ad was higher than that of the copywriter ad. Interestingly, through a crowdsourcing experiment, we show that consumers judge health ads written by copywriters as more persuasive than ads from other sources (algorithm and users). These results suggest that algorithms can be trained to write health ads that are, in practice, as good as those of copywriters.
Language: English
Subfield: Sociology and Political Science
Field: Social Sciences
Domain: Social Sciences
Open Access status: gold
OpenAlex ID: W4411549795