Taboola study finds: AI ads match human-made campaigns at scale

A new academic study examining more than 500 million digital ad impressions suggests that advertisements created using generative artificial intelligence perform at levels comparable to those developed by humans.

The research, conducted using live campaign data from Taboola’s performance advertising platform Realize, was carried out in collaboration with researchers from Columbia University, Harvard University, the Technical University of Munich and Carnegie Mellon University. The study analysed hundreds of thousands of ads that collectively generated over three million clicks.

The findings indicate that AI-generated ads achieved similar click-through rates to human-made ads once statistical controls were applied. In the raw data, AI ads recorded a marginally higher average click-through rate of 0.76 per cent compared with 0.65 per cent for human-created ads.

One of the key observations from the research was that AI-generated ads performed best when they did not appear overtly artificial. Ads perceived as more human-like generated higher engagement than both clearly AI-styled creatives and traditional human-made ads.

The presence of large, clearly visible human faces was identified as a significant trust signal across campaigns. According to the study, AI-generated ads were more likely to include such visual cues than their human-created counterparts, partly due to platform guidelines and creative best practices.

The research also found that using AI for creative production did not negatively affect downstream conversion performance, suggesting that advertisers can scale production without sacrificing effectiveness. Early adoption of AI-generated ads was most prominent in categories such as food and drink and personal finance.

“Taboola’s platform provided us with a literal gold mine of real-world data that is simply unavailable in a lab setting. By analysing over 500 million impressions, we were able to move past the hype of GenAI and uncover its real impact in large scale settings,” said Oded Netzer, Vice Dean for Research, Columbia Business School. “Our findings prove that when AI is used to enhance human cues—like the trust found in a human face—it doesn't just match human performance; it often sets a new ceiling for engagement.”

The study used a quasi-experimental methodology known as “sibling ads”, comparing AI-generated and human-made creatives produced by the same advertiser for the same campaign on the same day. This approach helped isolate the impact of creative origin while controlling for variables such as audience targeting, timing and landing pages.