Fake news is often driven by deeply ingrained partisanship. Its proliferation is also fueled by financial motivation, of course. New research, Market Forces: Quantifying the Role of Top Credible Ad Servers in the Fake News Ecosystem, found that that the top-10 credible ad servers alone account for 67% of fake news and 56% of and low-quality ad traffic. Further, Google delivers 48% of the ad traffic on fake news publishers and 32% on low credibility sites.

Methodology

To conduct this research, Lia Bozarth and Ceren Budak at the University of Michigan used a web API tool to simulate peoples’ browsing behavior. Their research identified fake news sites as those filled with pseudoscience and false facts. They also found that low credibility publishers are often hyper-partisan. The API tool initiates a browser session, navigate to the site’s homepage, scrolls through the page, and focuses the mouse on each ad’s iframe. Bozarth and Budak then filtered the collected data by the top 50 ad serves. In all, 84% of their ad data was served by the top 50% ad servers.

This methodology allowed the authors to measure ad server usage of high-quality sites, fake, and low-quality news sites. Three-quarters (74%) of all publishers (high-quality, fake and low-quality) have one or more ad servers displaying ads and 26% of all publishers are ad-free. The findings also show that fake news publishers often have fewer ad servers on average compared to low-quality and traditional news sites. Further, the median number of ad servers for fake and traditional sites is six and eight ad servers, respectively.

Ad server quality

Consumers navigating to fake and low-quality news site are at potential privacy and security risks. Fake and low-quality news publishes have a higher tendency to serve more ads and to partner with riskier ad servers when compared to traditional news media with similar popularity and age skew. Importantly, while fake news publishers use riskier ad servers, they are also dependent on credible ones. According to the findings, the top credible ad servers play a substantial role in delivering ad revenue to fake news sites. For instance, 6.7% of all fake domains are dependent on top-10 ad servers.

Impact on ad server revenue

Bozarth and Budak also provide a weighted ad traffic analysis to illustrate the role top credible ad servers play in providing revenue to fake and low-quality news sites. They found that full 61% of fake news ad revenue is estimated to be supplied by the 10 credible ad servers.

In terms of revenue to the ad servers, Bozarth and Bodak estimate that the top-10 firms, in aggregate, generate $24,500 to $28,600 monthly through fake news sites. That’s an average of $985,700 to $1.15 million of revenue or approximately 0.1% to 1.0% of the top 10 ad servers’ their annual revenue.

While the authors support ad servers blacklisting fake and low-quality news sites, that may not be the best answer. Profit-driven fake news sites banned by top-tier ad servers will partner with less reputable ones without hesitation. Further, owners of blacklisted websites will also migrate to new domains. It becomes a game of “whack-a-mole” unless there is an ability to detect these sites before ads are served. Additional research is needed to clearly identify fake and low-quality news sites. Criteria like the type of advertisers on these sites and their CPM or CPC rates to identifying profit models by category sectors. Each additional factor identified as a predictor of fake and low-quality news site can help inform ad servers about the content qualify of news sites. And this, in turn, supports sites that provide better consumer experiences and higher quality information.