In digital advertising, marketers often distinguish between premium and nonpremium websites. In programmatic advertising, the risk that ads will (often unintendedly) end up on nonpremium websites is typically higher than with direct deals. The question is how bad is it when advertisers appear on nonpremium websites? The answer is: It depends on campaign goal (branding or performance) and the brand's premiumness itself. For premium brands (e.g., BMW), we don't find any recall effects, but we do find adverse effects of nonpremium advertising on attitudes and clicks. For nonpremium brands (e.g., Dacia), we again see no effects on recall. Attitudes are also not negatively affected when advertising on nonpremium websites, but clicks do suffer (even though not as much as for premium brands).

So why are these findings important for programmatic advertising? Programmatic is more prone to brand safety issues than direct deals because ads bought via programmatic might unintendedly end up on websites of low quality. The New York Times reported that global players have unwantedly been subsidizing Russian government-run propaganda websites (e.g., RT.com) through programmatic. Websites such as RT.com attracted ads from 477 companies, including Amazon, PayPal, and Walmart. Our findings suggest that marketers can use programmatic for branding campaigns of nonpremium brands as well as campaigns aiming to increase recall for premium brands. For performance campaigns, programmatic is only advised if the cost benefits outweigh the negative click effect.

In terms of future research, it would be super interesting to see whether these negative website quality effects can be offset by targeting. In addition, following a nonpremium strategy might still be profitable if CPMs for nonpremium sites are substantially lower.

The paper provides new evidence that marketers should care about where their digital ads appear, which has implications for their use of programmatic advertising.