A new research paper that’s been getting a lot of media attention argues that Google and Facebook owe news publishers $14 billion a year “for the use of news content if the Journalism Competition and Preservation Act (JCPA) comes into force.” We’ve already seen that number used in advocacy for the JCPA’s passage. So we decided to dig into the paper. We found that stacking methodologies from past sources and studies significantly overstates the estimates – and represents the wrong approach to news policy.
As regular readers know, we spend a lot of time advocating for policies to support local news. We even authored our own proposal to create a fund for news based on contributions from dominant digital platforms. We also strongly advocate for policies designed to break down the power of Big Tech. But we have consistently joined with other civil society organizations and small and minority-owned media to advocate against the JCPA. For more on why, view our coverage of the bill.
The Latest Salvo in the Battle of Data
It’s important to note that the new research paper is only the latest in a volley of analyses by news organizations and their lobbyists designed to claim the high ground in the estimate of value exchanged by publishers and platforms. For example, in 2018, News Media Alliance hired Keystone Strategy, a consulting firm, to estimate Google’s revenue from news content at $4.7 billion. The analysis came under significant criticism. One business analytics professional pointed out that the entire methodology was based on a Google executive’s offhand comment at a conference – 10 years earlier. Prominent journalists and academics described it as “humiliating,” “bogus,” and based on “very little data.” Harvard’s Nieman Lab described the estimate as “imaginary.” And a prominent academic focused on the news industry called it “fuzzy.” In 2021, the National Association of Broadcasters commissioned another consulting firm to estimate that broadcasters lose close to $2 billion per year in value their news content generates for Google and Facebook. Facebook’s News Feed alone was estimated to create almost $500 million in lost value for broadcasters – a hard number to believe, especially now that Facebook has decided to wind down the News Feed in many countries.
Facebook and Google, of course, have rebutted these analyses with data of their own. Facebook has published its commissioned research showing that news stories make up less than three percent of what shows up in users’ feeds. And Google did its best to debunk the new research paper, stating that “less than two percent of all searches are news related and we don’t run ads or make money on the vast majority of them.” Both platforms cite advertisers’ reluctance to advertise adjacent to what might be controversial or polarizing news content – a dynamic publishers have complained about, as well.
Breaking Down That $14 Billion Estimate
The authors of the new working paper – respected researchers from The Brattle Group, the University of Houston, and Columbia University – begin by describing the limits of their methodology (though – cough – we haven’t seen any of these limits in the news reports). In fact, they acknowledge that none of the best three methodologies that researchers would ideally use to calculate the share of the platforms’ advertising profits that come from news content are feasible because the platforms won’t provide the data required to apply them. (We agree this is a problem, and think platforms should be required to make more data available for researchers, such as is called for in the PATA Act.)
The authors also acknowledge that their methodology is based solely on impressions of news content on the platforms, ignoring any click-through by platform users to the actual news article. That means the methodology, already compromised from the ideal, deliberately ignores the value of any traffic sent to the news outlet where it can be monetized by the publisher. That is the first way in which the estimates of who owes what to whom are overstated: They disregard the value of customer acquisition to publishers. They pay little attention to the natural experiment that’s happening in Canada, which shows that Meta’s blocking of news content has created devastating losses of traffic to news publishers (especially small ones) with no loss of engagement for Facebook, Meta’s market-dominant social media platform. The authors also disregard existing payments to publishers as being too “difficult to estimate in the U.S. reliably.”
The actual calculations of the value of news content to Google rely largely on a 2023 study by Fehr Advice and Partners commissioned by news publishers in Switzerland. The authors of the Fehr Advice study reference a 2022 book that included statistics showing that 50-60 percent of internet searches (the U.S. authors lock on to 50 percent) are “information searches” (rather than transaction or navigation searches). Because of their “quality, completeness, and trustworthiness,” the study’s authors argue that all of these “information searches” entail news results. (Apparently, the authors have never heard of Wikipedia, one of many online sources of information and one which often appears at the top of information searches.) This is another source of overstatement. Fehr Advice also conducted an online behavioral experiment that culminated in asking respondents whether they would “prefer” to see search results with or without news content in them: 70 percent of respondents reported that they would prefer search results with news. But in their calculations, the U.S. researchers translated this stated preference among users into actual demand among searches. That is, they assumed that 70 percent of all searches that include news results – that is, all “information searches” – are the result of “demand” for news content. This is likely a significant overcall on what proportion of searches specifically call for news content.
In other words, the U.S. researchers multiply these two estimates:
The 50 percent of all searches that are information searches
(deemed “news searches”)
The 70 percent of research respondents that said they’d prefer to see news in their search
(deemed “demand for news” in searches)
35 percent of “all demand for searches in the U.S. is from users seeking news media content”
Further, despite agreement among platforms and publishers that advertisers are less likely to advertise on news content, the authors assume that advertising occurs at an equal rate and price on these searches as on, for example, transaction searches that actually lead to a purchase. This is another source of overstatement, and it leads to the highly unlikely conclusion that 35 percent of Google’s total U.S. ad revenue is associated with news content.
These spoils then need to be split between the platforms and publishers somehow. The authors refer to “insights from the economics of bargaining” to justify the use of a 50/50 split of the estimated news-based advertising revenue between publishers and platforms. Et voila: $10-12 billion. The “insights” come down to this: One possible outcome of any two-party negotiation is 50/50. There’s no evidence of why this would be the most likely, or even the fairest, outcome in this negotiation. In fact, the six historical benchmarks they reference as support are all from traditional media licensing agreements – from 15 years ago.
The calculations of what Facebook “owes” publishers use a similarly layered approach. The authors use a series of Facebook financial filings, industry advertising market data, and a different study about news consumption habits to estimate the share of time each respondent spent on Facebook consuming news. Again, all the data on news consumption was self-reported. This resulted in an estimate of Facebook’s “ad revenue from news consumption” – and again, despite agreement among platforms and publishers that this is less conducive to advertising, it was assumed that ads occurred at the same rate on news as other forms of content. The researchers then again assigned publishers 50 percent of it: $2 billion. This division is subject to the same caveats as the one for Google.
Let’s end where we began: We agree Google and Facebook have too much market power in digital advertising. We agree there should be more transparency about the existing agreements between platforms and publishers and more platform data available to researchers. We have advocated for various proposals, including our own, to fund local journalism using Big Tech money, and we have supported litigation and legislation to break up the power of these advertising duopolies. But enough of the battling data designed to support a fundamentally flawed proposition. The Journalism Competition and Preservation Act is the wrong solution, and trying to estimate the unjust enrichment of platforms is the wrong approach. We need policy solutions based on public interest obligations, as my colleague Harold Feld has written, and the incalculable value of news to the public.