The Winners and Losers in a World of Digital Advertising

What began in 2007 with Facebook Ads—which allowed advertisers to reach users based on data pulled from their Facebook profiles—has morphed into a $300-billion digital ad industry ruled by Google, Meta, Amazon, and TikTok. It hoovers up users’ profile and online activity data and uses AI and machine learning to automatically send them ads exquisitely targeted to achieve the most clicks and purchases.

When Ads Help—and Harm—Consumers

When a new technology like digital advertising suffuses the human experience—adults, on average spend more than three hours per day on social media, and even more on streaming platforms—it’s worthwhile to step back and try to take stock of the impacts. That’s the spirit behind a new working paper written by Tuck professor James Siderius and co-authors from MIT: Daron Acemoglu, Daniel Huttenlocher, and Asuman Ozdaglar. 

In “Online Business Models, Digital Ads, and User Welfare,” the researchers begin from the solid premise that digital ads can be both angel and devil. Ads can be beneficial by informing people about a product they might want to use. But they can also be enticing and manipulative, exploiting users’ biases and vulnerabilities, causing them to make purchases they later regret. The latter category, for example, has included ads for dieting apps and “detox teas” targeted at teen girls, reinforcing harmful beauty standards, and payday loan companies targeting low-income users and people searching for financial help. 

To understand the hard logic of a two-sided platform that collects ad fees from firms, and offers entertaining content for users, Siderius and his co-authors build a model of the real world. In the model, the platform “can monetize its services via advertising, subscription fees, or both,” they explain. “Digital ads are informative about the (user-specific) quality of the firm’s product. Ads are therefore beneficial because they provide informative signals, but are also costly because they interrupt the entertaining content.” But the model then goes further and simulates the diversity of the users on the other end of this content. It has “sophisticated” users who internalize the persuasive power of digital ads when deciding whether to purchase, and “naïve” users who take the ads at face value and underestimate the likelihood they are being preyed upon. “It’s not that sophisticates are better at differentiating ‘good’ from ‘bad,’” Siderius explains, “but they understand the technology is intentionally targeted and so respond to that information in a more measured way.” 

What the Research Reveals

At the outset, it was not clear to the researchers how the balance would tip, given the dual nature of ads and the people who potentially see them. But when they ran the model, the ambiguity faded. “We found that across the board, based on how platforms set their subscription prices, how many ads they choose to show, and the kind of targeting technology they employ, welfare will always be worse for the users than if the platform were to adopt a pure subscription model,” Siderius says. Welfare degrades on two levels. The ads interrupt the content, thereby harming the viewing experience. And the increased purchasing by naïve users drives up prices for everyone. “That was the first surprising result,” Siderius notes.

The second? That competition among firms and platforms didn’t mitigate the negative effects of digital ads. The researchers had surmised that competition would lead, as it usually does, to lower prices. But the confounding factor here is the ad targeting. The technology to target users is so powerful that it allows firms to not only reach individuals but also express how superior they are to their competitors. This sets in motion the gears of the market: digital ads become more valuable to firms, which means they pay more for them, which means the platform is going to over-advertise to users, relative to a normal benchmark of informativeness. It all leads to increased profits for firms and platforms, an annoying viewing experience, and greater manipulation of naïve users. “When you think about a general equilibrium model where you have the platform setting the business model and maximizing profits, and firms doing the same thing,” Siderius says, “what we find is actually this manipulation channel wins out, and the users are worse off.”

Toward a Fairer Digital Marketplace

There might be a way to prevent this outcome. Following the economic principle of “tax bads, not goods,” which advocates for taxing harmful activities to discourage their use, the co-authors investigate the effect of a flat, digital ad tax on platforms. In this scenario, the government would set a tax rate for the portion of platforms’ revenues derived from digital ads. Lo and behold, the researchers find that such a tax would improve outcomes and get platforms to transition away from digital ads to subscriptions. 

Given the political clout of platforms today, a digital ad tax may not be in the immediate offing. But Siderius says there’s another way to reform the system: digital literacy programs that teach people how to be more discerning customers. And not everyone must become more sophisticated. If more people are careful about how they respond to digital ads, it disciplines the platforms to use ads more responsibly, and less manipulation will result.

For those of us wanting to join the sophisticated user group, Siderius offers some advice: “Be aware the ads are targeted at you and showing you products and messages designed to get you to purchase. Because of this, you should comparison shop, leave the platform to do your own research, and take the time to make an informed purchase decision.”