November 4, 2016 – The text of this article is below, or click here to access our library.
Facebook pulls a number of levers to keep users on its own platform rather than going to the websites of publishers who fuel Facebook with free content. Such tactics mirror conduct that has landed Google in antitrust trouble in Europe. As with the allegations against Google, Facebook appears to lack algorithm neutrality, prioritize content that benefits the platform, and discriminate against competitors.
Facebook competes with news publishers for users’ time spent online, user data, and online advertising dollars. Social news referral, news distribution, user data, online advertising, and social media are all relevant markets in which antitrust regulators may scrutinize Facebook. In this article, we discuss the antitrust risk created by Facebook’s vertical integration into publishing media content on its own platform with Facebook Instant Articles (FBIA) and in turn prioritizing such content in its news feed.
Antitrust and Political Scrutiny of Facebook
Although Google has dominated headlines in Europe, Facebook is under antitrust scrutiny as well. In March, the German Cartel Office opened an investigation into Facebook’s data collection practices and terms and conditions. Privacy law and antitrust law are converging in Europe, and Commissioner Vestager has stated that big data creates concerns regarding abuse of dominance. The German data protection Commissioner also ordered Facebook in September to stop collecting data on WhatsApp users.
In the U.S., unprecedented control over the lens through which citizens view the news could draw federal scrutiny. Concentration of news distribution has historically been a point of interest for federal lawmakers and regulators. Recently, conservative lawmakers questioned the role of Facebook in presenting biased news results. For instance, Senator John Thune (R - South Dakota) sent a letter in May about the company's role in suppressing conservative viewpoints in its "Trending Topics" section. Political pressure caused Facebook to change its Trending Topics practices, but this issue shows that platform neutrality is of political concern in the U.S. as well.
Last month, Caleb Gardner, former social media adviser to President Obama, spoke about the phenomenon of “filter bubbles,” at a speech at Northwestern University. “More likely than not, you get your news from Facebook,” he said. “Forty-four per cent of US adults get news on the site, and 61% of millennials … if that doesn’t frighten you, you don’t know enough about Facebook’s algorithm. If you have a parent who’s a Trump supporter, they are seeing a completely different set of news items than you are.”
In-depth Look at Facebook’s Market Power in News Distribution and Advertising
Facebook’s platform has supplanted direct news sources as the primary mechanism for most news distribution. According to statistics provided by Parse.ly, Facebook controls around 45% of all online news distribution and around 80% of all social news distribution. Sixty-six percent of US Facebook’s 1.71 billion users receive news from the platform. 72% of adult American internet users use Facebook, according to Pew Research. They spend an average of 50 minutes on the platform daily, “And time enables Facebook to learn more about its users — their habits and interests — and thus better target its ads,” reports the New York Times.
Facebook and Google have a duopoly in digital advertising. In the first quarter of 2016, 85 cents of every new dollar spent in online advertising will go to Google or Facebook, according to The New York Times. In terms of functionality, Facebook and Google use news websites for the benefit of their platforms. Both have dominant positions on the two alternate sides of traffic referral – Facebook dominates social referral with an 80% share while Google dominates search referral with an 85% share, according to statistics from analytics provider Parse.ly from June through August.
In-depth Look at How Facebook Favors Native Content
Rather than directing traffic to publishers’ websites, the longstanding rationale for publishers providing free content to platforms like Facebook and Google, FBIA retain users on Facebook’s own platform. Facebook retains highly granular data on users, growing its existing market power in the Facebook/Google digital advertising duopoly.
Publishers give up highly granular data collected by cookies on their own websites in exchange for less insightful data shared through Facebook’s platform, the accuracy of which publishers cannot verify. Most publishers and industry participants we spoke with agreed that using Facebook’s native content channels was an arrangement that heavily favored Facebook. Despite that, several aspects of Facebook’s platform push publishers into making use of Facebook’s native channels.
Facebook’s in-app browser degrades user experience for non-native content. The main consideration for using Instant Articles is faster load times for content. Facebook itself uses load times as the main feature in marketing for the service, and the company has noted in its literature and during developer events that Instant Articles load 10 times faster than mobile articles on a standard browser. Facebook claims to have solved a problem with the mobile Internet at large through hosting content on its own platform, with the cost for publishers being significant audience data and 30% of any ads that publishers cede to the Facebook Audience Network.
Facebook, however, has not provided a second option where only there was one previously. Instead, the company has worked to disable the ability to automatically load articles in vanilla Safari or another browser outside of the app in iOS rather than Facebook’s in-app browser, a feature which the app used to have. Further, it appears that the in-app browser actually works to create some of the friction that Facebook says Instant Articles is meant to alleviate, which prods publishers into utilizing Instant Articles. The Facebook in-app browser is a version of Safari that Apple makes available for developers to customize for their own purposes. It appears that either Facebook has not maintained its in-app browser at the same pace as other browser developers or it adds some processes overtop of the simple API that serves to increase load times, though Facebook could not be reached for comment.
Some publishers told us Facebook’s in-app browser is essentially unusable at times due to load times, making Instant Articles the only viable option. Others said the experience through the in-app browser is so degraded, they feel users will look at their website or content in a negative light, rather than Facebook, and judge their brands. We found in a test of 24 articles loaded on both the vanilla Safari app on iOS and the iOS Facebook in-app browser, the Facebook in-app browser apps loaded more than four seconds slower on average than Safari.
Facebook prioritizes its native content in the News Feed. Another pressure for publishers to use Instant Articles is the view that Instant Articles are prioritized within user News Feeds, thereby gaining more exposure and user interaction. Although Facebook denies that its News Feed algorithms directly prioritizes Instant Articles through its ranking system, Instant Articles are “naturally” prioritized and appear higher within News Feeds than non-Instant Articles content because their faster load times increase “interactions.”
Interactions are the user-based cues – Likes, Reactions, Comments, and Shares – and the more interactions a piece of content on the platform has, the more likely it is to appear higher in someone’s News Feed. According to Facebook, users click on Instant Articles 20% more than other articles and share Instant Articles 30% more than mobile web articles on average, “amplifying the reach of your Stories in News Feed.” Essentially, the company states it is not prioritizing the content and then uses the prioritization of the content as a selling point in marketing the service. One anecdotal comparison between an average mobile article and an Instant Article posted from Hootsuite found the article saw a 37% increase in reach, 22% increase in Shares, 83% increase in Likes, and 57% increase in link clicks.
Facebook’s prioritization of native content mirrors that of Google prioritizing its own sites in the EU Google Shopping Case. The European Commission’s Statement of Objections in the Google Shopping case alleged that, starting in 2008, “Google systematically positions and prominently displays its comparison shopping service in its general search results pages, irrespective of its merits.” To remedy its concerns, the Commission is expected to seek a non-discrimination standard, which is likely to include the following requirements:
• Positioning, ranking, labeling, and displaying competitors’ content in the same way as Google’s own;
• Sharing access and space on pages with competitors in a clear way that does not deceive or mislead consumers;
• Using an unfettered algorithm to generate organic search results in its shopping service;
• Applying the same system of penalties to its own service as it does to competitors.
Google’s conduct regarding Shopping is not only problematic under European Antitrust law. As the leaked FTC memorandum on the Google investigation showed, staff took issue under U.S. antitrust law with Google’s demotion of competing vertical websites in its search results. Here, Facebook similarly is demoting content housed on publishers’ own websites in favor of the native FBIA. Facebook competes with news publishers for users’ eyeballs, time spent online, and data. Thus, its newsfeed is prioritizing native content to the detriment of competitors. The EU could similarly view the demotion of non-native content as and abuse of dominance and seek a non-discrimination standard comparable to the one it is expected to impose on Google.
In-Depth Look at How FBIA Deprives Publishers of Big Data
In digital publishing, audience data is the competitive factor. For those in the digital publishing space, the competitive differentiator between publishers is audience data. Paul Niwa, chair of the department of journalism at Emerson College, has helped launch several media outlets throughout his career. By his view, the fundamentals of publishing and monetizing an audience have not changed much from the heyday of the newspaper. “It’s not only being able to tailor what content you want to produce for that audience to engage them, but it’s also in how you work with advertisers to be able to present a certain type of audience to an advertiser and sell access to that audience. The ability to track, to be able to gather information, to build a relationship beyond just readership or viewership is really critical,” Niwa said.
Older processes to engage that audience – like circulation data for newspapers – have, however, adapted to the digital world along with the content. For one, certain data allows publishers to thoroughly analyze the performance of pieces of content and determine how to better produce in the future. That involves metrics like views, time spent on page, click-throughs, bounce rates, and other indicators. When publishers review these numbers, they can see what works and build strategies to better serve their audiences and optimize campaign performance.
More crucial than readership numbers are audience profiles used to attract advertising. The most valuable asset for a media company when it comes to generating advertising dollars is well-vetted and rich audience data that shows the profiles of users reading the content. Advertisers are more likely to run with publications that can prove they have specialized audiences or subsets of audiences who are highly engaged with content and receptive to relevant advertising campaigns. Publishers amass this type of data through third-party services, analytics tools, and other means.
The most important and primary source, though, are first-party cookies. These bits of data are stored on the user’s browser when he or she visits a website and can read a user’s web history and other information to build a profile for the user that the publisher can group and sell. Cookies are updated on return visits, allowing publishers to compound data sets and build out existing profiles.
Importance of first-party cookies. A non-FBIA link within the Facebook app, like any other news aggregator redirect, will link to the publisher’s website. Once the user is on a website that the publisher controls, the publisher can apply any number of processes to glean audience data from the user, with the most important and proprietary being first-party cookies. Below is an example of 28 first-party cookies applied by Bloomberg upon visiting its website (with a handful of others not pictured from various other Bloomberg-controlled domains):