If we want to solve the disinformation crisis, we need to fix the perverse incentives created by the online advertising market through structural and regulatory solutions.
With candidates already beginning to plod the campaign trail and participate in debates for the 2020 US presidential election, tech companies are gearing up to battle the scourge of viral disinformation. But while action on the part of individual platforms may help to address some of the issues that became notorious after the 2016 election, the problem of information disorder—or at least the portion of it that’s motivated by profit—has partly to do with the perverse incentives created by the online advertising market, which call for broader structural and regulatory solutions.
In the wake of the 2016 election, journalists uncovered numerous examples of individuals and groups making tens of thousands of dollars in ad revenue from hoax news stories. In a study published this year, Jessica Eklund and I set out to interview executives and employees in the adtech industry, to see how they viewed the issue as well as its potential solutions. What we found was that one of the dominant forms of online advertising—programmatic advertising—is structured in a way that makes addressing problems particularly difficult.
Garden-variety programmatic advertising is also called “real-time bidding” or RTB. In this form, publishers offer up ad space for sale on their websites and advertisers bid to place their ads in that space, based largely on what they know about you, the user loading the page. A collection of intermediaries, adtech firms, handle the details of each such transaction. By the time your web page has finished loading, this fully-automated auction has already completed, and the winning advertiser’s ad appears in your browser. Similar versions of this process exist for placing ads on mobile and smart TV apps as well.
One of the most important things to note about the RTB process is the number of intermediaries involved in it. A fairly straightforward ad buy involves three to four intermediaries standing between the brand and the publisher. In the middle of this supply path is the ad exchange, the digital auction house in which ad space is actually sold. But the ad exchange is only one link in the chain. To connect to the ad exchange, advertisers typically go through an intermediary service called a supply-side platform (SSP), and publishers similarly use a demand-side platform (DSP) to manage their respective inventories of ads and ad space.
Beyond the four intermediaries illustrated here (agency, SSP, exchange, DSP), additional companies may also be parties to the transaction. For example, many transactions also involve data-management platforms (DMPs), which warehouse data advertisers employ to profile web users. And much as in the high-speed trading world from which RTB takes inspiration, other firms will buy and resell ad space in the exchanges solely to make a profit from flipping inventory, adding even more complexity to the supply path in the process.
Having this many links in the chain between the advertiser and the publisher can dramatically decrease the level of transparency brands have regarding where exactly their ads are appearing on the Web, which is exactly what opportunistic publishers of hoax news stories and miracle-diet pages take advantage of when they offer up ad space for sale on dubious sites.
Moreover, as Jason Kint—head of the publisher trade group, Digital Content Next—pointed out to us, many of these intermediaries aren’t dealing directly with any of the primary stakeholders in the transaction. Firms ensconced comfortably in the middle of the supply chain are making money from each ad sold but aren’t immediately answerable to advertisers, publishers, or users. This creates incentives for all sorts of less-than-ideal practices on the part of these adtech firms, ranging from merely looking the other way when questionable sites bid on ads to ignoring actual fraud perpetrated by shady publishers to, in the most egregious cases, actively engaging in fraud themselves, ginning up and selling fake traffic to advertisers.
Because there are so many parties to each transaction, adtech firms are often able to diffuse responsibility for poor ad placements, gesturing toward other services in the supply chain to explain problems—another vendor that improperly classified a site’s content, say, or allowed a bad actor into the transaction. This seems to lead to a bystander effect, wherein the emergent issues are seen as someone else’s problem.
An additional factor leading to the monetization of disinformation through ad dollars has been the way in which RTB incentivizes brands to focus on reaching desirable users at the best price to the exclusion of concern for editorial context. If you sell shoes and the behavioral targeting data you’ve collected on user x suggests they’re in the market for a new pair of sneakers, it’s nearly always cheaper to place an ad in front of that user while they’re visiting a hoax news article than while they’re reading the New Haven Register.
Moreover, if the advertiser assumes user x is interested in the hoax news article, there’s little risk of damage to the brand if the company places its ad there. In other words, behavioral targeting as an advertising strategy risks creating a race to the bottom in terms of the quality of the content monetized by advertisers.
Fortunately, the 2016 presidential election highlighted the profits made from hoax news sites. Coupled with the efforts of activist groups like Sleeping Giants to encourage advertisers to pull their advertising from publications like Breitbart News, Nyheter Idag, and Boulevard Voltaire, whose headlines often toe or cross the line into hate speech, the effect has been to increase concern among advertisers over where their ads appear online and what sorts of content they monetize.
High levels of fraud—along with simple ineptitude—in the online advertising marketplace have also yielded poor results for advertisers (some campaigns found that only 20 percent of their ads were viewed by actual audiences, as opposed to bots and click workers). Moreover, with so many parties to every programmatic transaction, even when things are running smoothly, intermediaries commonly skim somewhere between 55 and 70 percent of every dollar brands spend on RTB advertising. This has caused major advertisers like P&G to be increasingly selective about the firms and publications with which they spend their money. There is also a movement among advertisers, called “supply-path optimization,” to execute online advertising campaigns in ways that involve fewer intermediaries.
While I have argued here and elsewhere that the behavioral targeting paradigm in adtech impacts deeply on the health of our news and information ecosystem, this world is more often discussed in terms of digital privacy. Thankfully, privacy reforms—starting in Europe—may have ancillary benefits for legitimate news publishers. For example, the Information Commissioner’s Office (ICO) in the UK recently published a document arguing that the opacity in the RTB supply chain, combined with the number of parties engaged in each programmatic transaction, creates dire user-consent issues that effectively put the whole RTB advertising ecosystem in breach of the GDPR. After all, these interactions all involve behavioral targeting and therefore some degree of access to users’ personal data.
Between industry trends toward supply-path optimization, pressures levied by the GDPR, and the fact that the programmatic ecosystem includes hundreds—if not thousands—more adtech companies than most industry observers believe it can ultimately support, it seems clear that there will be some level of consolidation. During our study, we heard from an executive at one large adtech firm that the transparency necessary to give advertisers more control over where their ads appear would ultimately be achieved through vertical integration. If one large tech company operated all the points in the supply path, the argument went, there would be greater visibility and accountability across all the steps in the ad placement process.
|“Regulators should be aware of how behavioral targeting not only impacts privacy but—within the current market structure—incentivizes the creation of disinformation.”|
Of course, as you’re probably thinking, the online advertising space is already dominated by the duopoly of Google and Facebook. Google owns firms at many points along the programmatic supply chain, but was nonetheless described in 2016 as the “financial engine” behind fake news by trade journalist Craig Silverman. Meanwhile, Facebook owns a platform through which advertisers reach consumers with no other intermediaries involved. As Sally Hubbard has argued, the fact that disinformation has flourished on Facebook should give us all pause with respect to the idea that unchecked industry consolidation will reduce our problems with disinformation and “fake news.”
What is clearly needed—given all the disincentives to self-reform within the adtech industry—is some level of careful regulation in this space, demanding that a programmatic ecosystem in which numerous firms operate is not only competitive but also responsible. Alternatively, regulations could ensure that emergent monopolies are well regulated and required to support a healthy information ecosystem through civically responsible business practices and, perhaps, various forms of offsets to restore lost revenues to public-interest media.
Laws such as the updated Loi Sapin in France, aimed at increasing transparency and limiting corruption within the advertising industry, could provide a model for some aspects of these reforms (though not all). Brave Software’s Policy and Industry Relations lead, Johnny Ryan—whose complaints led, in part, to the recent ICO report—has argued that the amount of data collected in the service of behavioral targeting should be dramatically scaled back. And, it should be noted, this position is well supported by empirical data. In a recent study, for example, researchers Veronica Marotta, Vibhanshu Abhishek, and Alessandro Acquisti showed that the revenue benefit to publishers of enabling user tracking may be as low as 4 percent.
Overall, regulators should be aware of how behavioral targeting not only impacts privacy but—within the current market structure—incentivizes the creation of disinformation. They should balance the issues of opacity and low accountability that seem to attach to having numerous firms involved in each RTB transaction with the ills created by industry consolidation. As we’ve seen, monopolists tend not to behave in the public interest, while the smaller independent firms left competing for their scraps seem more likely to cut corners, not less, as they struggle to stay afloat.
Joshua Braun is an Associate Professor of Journalism at the University of Massachusetts Amherst. The author would like to thank Sally Broughton-Micova for her formative comments on the intersections between our study and larger policy conversations around digital advertising.
The ProMarket blog is dedicated to discussing how competition tends to be subverted by special interests. The posts represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty. For more information, please visit ProMarket Blog Policy.