Tag Archives: facebook ads

Apps, News, Observation, Politics, Social Media, Tech

Facebook Isn’t Telling You How Popular Right-Wing Content Is on the Platform

November 18, 2021 Creative Commons

Facebook insists that mainstream news sites perform the best on its platform. But by other measures, sensationalist, partisan content reigns

In early November, Facebook published its Q3 Widely Viewed Content Report, the second in a series meant to rebut critics who said that its algorithms were boosting extremist and sensational content. The report declared that, among other things, the most popular informational content on Facebook came from sources like UNICEF, ABC News, or the CDC.

But data collected by The Markup suggests that, on the contrary, sensationalist news or viral content with little original reporting performs just as well as—and often better than—many mainstream sources when it comes to how often it’s seen by platform users.

Data from The Markup’s Citizen Browser project shows that during the period from July 1 to Sept. 30, 2021, outlets like The Daily Wire, The Western Journal, and BuzzFeed’s viral content arm were among the top-viewed domains in our sample.

Citizen Browser is a national panel of paid Facebook users who automatically share their news feed data with The Markup.

To analyze the websites whose content performs the best on Facebook, we counted the total number of times that links from any domain appeared in our panelists’ news feeds—a metric known as “impressions”—over a three-month period (the same time covered by Facebook’s Q3 Widely Viewed Content Report). Facebook, by contrast, chose a different metric, calculating the “most-viewed” domains by tallying only the number of users who saw links, regardless of whether each user saw a link once or hundreds of times.

By our calculation, the top performing domains were those that surfaced in users’ feeds over and over—including some highly partisan, polarizing sites that effectively bombarded some Facebook users with content.

These findings chime with recent revelations from Facebook whistleblower Frances Haugen, who has repeatedly said the company has a tendency to cherry-pick statistics to release to the press and the public.

“They are very good at dancing with data,” Haugen told British lawmakers during a European tour.

When presented with The Markup’s findings and asked whether its own report’s statistics might be misleading or incomplete, Ariana Anthony, a spokesperson for Meta, Facebook’s parent company, said in an emailed statement, “The focus of the Widely Viewed Content Report is to show the content that is seen by the most people on Facebook, not the content that is posted most frequently. That said, we will continue to refine and improve these reports as we engage with academics, civil society groups, and researchers to identify the parts of these reports they find most valuable, which metrics need more context, and how we can best support greater understanding of content distribution on Facebook moving forward.”

Anthony did not directly respond to questions from The Markup on whether the company would release data on the total number of link views or the content that was seen most frequently on the platform.

The Battle Over Data

There are many ways to measure popularity on Facebook, and each tells a different story about the platform and what kind of content its algorithms favor.

For years, the startup CrowdTangle’s “engagement” metric—essentially measuring a combination of how many likes, comments, and other interactions any domain’s posts garner—has been the most publicly visible way of measuring popularity. Facebook bought CrowdTangle in 2016 and, according to reporting in The New York Times, has since largely tried to downplay data showing that ultra-conservative commentators like The Daily Wire’s Ben Shapiro produce the most engaged-with content on the platform.

Shortly after the end of the second quarter of this year, Facebook came out with its first transparency report, framed in the introduction as a way to “provide clarity” on “the most-viewed domains, links, Pages and posts on the platform during the quarter.” (More accurately, the Q2 report was the first publicly released transparency report, after a Q1 report was, The New York Times reported, suppressed for making the company look bad and only released later after details emerged.)

For the Q2 and Q3 reports, Facebook turned to a specific metric, known as “reach,” to quantify most-viewed domains. For any given domain, say youtube.com or twitter.com, reach represents the number of unique Facebook accounts that had at least one post containing a link to a tweet or a YouTube video in their news feeds during the quarter. On that basis, Facebook found that those domains, and other mainstream staples like Amazon, Spotify, and TikTok, had wide reach.

When applying this metric, The Markup found similar results in our Citizen Browser data, as detailed in depth in our methodology. But this calculation ignores a reality for a lot of Facebook users: bombardment with content from the same site.

Citizen Browser data shows, for instance, that from July through September of this year, articles from far-right news site Newsmax appeared in the feed of a 58-year-old woman in New Mexico 1,065 times—but under Facebook’s calculation of reach, this would count as one single unit. Similarly, a 37-year-old man in New Hampshire was shown 245 unique links to satirical posts from The Onion, which appeared in his feed more than 500 times—but again, he would have been counted just once by Facebook’s method.

When The Markup instead counted each appearance of a domain on a user’s feed during Q3—e.g., Newsmax as 1,065 instead of 1—we found that polarizing, partisan content jumped in the performance rankings. Indeed, the same trend is true of the domains in Facebook’s Q2 report, for which analysis can be found in our data repository on GitHub.

We found that outlets like The Daily Wire, BuzzFeed’s viral content arm, Fox News, and Yahoo News jumped in the popularity rankings when we used the impressions metric. Most striking, The Western Journal—which, similarly to The Daily Wire, does little original reporting and instead repackages stories to fit with right-wing narratives—improved its ranking by almost 200 places.

“To me these findings raise a number of questions,” said Jane Lytvynenko, senior research fellow at the Harvard Kennedy School Shorenstein Center.

“Was Facebook’s research genuine, or was it part of an attempt to change the narrative around top 10 lists that were previously put out? It matters a lot whether a person sees a link one time or if they see it 20 times, and to not account for that in a report, to me, is misleading,” Lytvynenko said.

Using a narrow range of data to gauge popularity is suspect, said Alixandra Barasch, associate professor of marketing at NYU’s Stern School of Business.

“It just goes against everything we teach and know about advertising to focus on one [metric] rather than the other,” she said.

In fact, when it comes to the core business model of selling space to advertisers, Facebook encourages them to consider yet another metric, “frequency”—how many times to show a post to each user on average—when trying to optimize brand messaging.

Data from Citizen Browser shows that domains seen with high frequency in the Facebook news feed are mostly news domains, since news websites tend to publish multiple articles over the course of a day or week. But Facebook’s own content report does not take this data into account.

“[This] clarifies the point that what we need is independent access for researchers to check the math,” said Justin Hendrix, co-author of a report on social media and polarization and editor at Tech Policy Press, after reviewing The Markup’s data.

This article was originally published on The Markup By: Corin Faife and was republished under the Creative Commons Attribution-NonCommercial-NoDerivatives license.

Check out Lynxotic on YouTube

Find books on Music, Movies & Entertainment and many other topics at our sister site: Cherrybooks on Bookshop.org

Enjoy Lynxotic at Apple News on your iPhone, iPad or Mac.

Lynxotic may receive a small commission based on any purchases made by following links from this page

Business, Observation

How The Daily Wire Uses Facebook’s Targeted Advertising to Build Its Brand

August 13, 2021 Creative Commons

Above: photo collage by Lyxotic

The social media giant’s powerful targeting tools appear to be part of Ben Shapiro’s success in growing his audience on the platform

Ben Shapiro, co-founder of The Daily Wire, a conservative media company, has mastered Facebook’s complex algorithms like no one else, posting links to stories from his publication that rank among the top 10 best performing posts on Facebook day after day after day.

What’s the key to his success?

As a recent NPR analysis shows, The Daily Wire’s sensationalist headlines garner a ton of engagement on a platform that rewards explosive content. But The Daily Wire is also a sophisticated user of Facebook’s advertising targeting tools to pinpoint users likely to be receptive to its outrage-driven brand of conservative content, The Markup has found.

Using data from our Citizen Browser project, we pulled targeting information from 241 Daily Wire ads that ran on Facebook between April 15 and July 15, 2021. We found that The Daily Wire largely chose to target people whom Facebook had pegged as interested in Fox News, Donald Trump, Rush Limbaugh, and other conservative mainstays, as well as individuals Facebook determined were characteristically or demographically similar to The Daily Wire’s existing audience members. (See our data here.)

Citizen Browser consists of a panel of roughly 1,800 Facebook users across the country who voluntarily share their Facebook news feed data with The Markup—providing a rare, albeit relatively small, window into what different people see on the platform.

By contrast, The New York Times—one of the largest legacy media publications in the U.S.—took a different tack in its Facebook advertising, targeting users according to the topics of the articles. So, for instance, an article about a band could be targeted to Facebook users with “music” listed in their ad interests. (Facebook says it determines users’ interests based on their past activities on the platform but has been somewhat cagey about how exactly this is done.)

Of the two publications, The Daily Wire used interest targeting more frequently than The New York Times did: 39.3 percent of Daily Wire ads versus 23.5 percent of ads from the Times were targeted in this way.

The table below shows the top 10 interests targeted in sponsored posts from both outlets:

While the Times mostly targets topical interests, of the top 20 interests targeted by The Daily Wire, only one (“American Football”) was not directly tied to conservative media or politics.

The Daily Wire also frequently made use of Facebook’s “lookalike audiences” feature to show content to new audiences of users who do not follow the page but share characteristics with those who do. In our dataset, 37.9 percent of Daily Wire posts used this type of targeting. The New York Times also used this targeting type, albeit rarely: Only 3.6 percent of its sponsored posts in our dataset targeted lookalike audiences.

“As you’re looking at this dataset, to me it shows that mainstream media outlets like The New York Times are still approaching the internet as a collective space in which you could potentially learn about anything, from ‘research’ or ‘science’ to ‘family and relationships,’ ” Francesca Tripodi, an assistant professor at UNC School of Information and Library Science at Chapel Hill, said. “But Daily Wire, if you’re saying, ‘We only want to target people who are interested in conservatism in America,’ that creates this bifurcated or dual internet, and that allows for information to circulate unchecked.”

Facebook advertising is designed to use personal data points about its users to guess what sorts of products they might like, she said, but there’s a fundamental difference between a food brand serving ads to people who like potato chips and a news brand serving information to people who like conservatism.

“[Daily Wire] is using the same tactics that these corporate entities are using but to create siloed interests around information,” Tripodi said.

Neither The Daily Wire nor Facebook responded to multiple requests for comment.

Beyond Facebook’s powerful data-gathering system, The Daily Wire amasses its own information on readers and potential readers.

The Markup also scanned Daily Wire ads in the Facebook ad library, which contains a broader range of ads than those seen by Citizen Browser panelists but does not disclose targeting information. Over a three-month period, from May through July, the ad library displayed 47 unique ads from The Daily Wire. Of these, 22 were survey-style ads prompting users to respond to emotive political questions.

Clicking the ad takes users away from Facebook and onto the dailywire.com domain, where they are asked to enter an email address in order to respond.

Over the same time period, no New York Times ads available in the ad library used this technique.

The Daily Wire’s website also contains an unusually high number of data-gathering trackers.

A scan from Blacklight, a website privacy inspector built by The Markup, on Aug. 4, 2021, turned up 41 ad trackers and 117 third-party cookies on the homepage. By contrast, The Markup’s scan of 100,000 of the most popular websites in September 2020 found an average of seven ad trackers and only three third-party cookies per site.

The site also uses Facebook’s bespoke Pixel tracking code to send data back to the social platform about users who have visited the site, which The Daily Wire can use to further tweak ad targeting and build new lookalike audiences.

“What you’ve shown here is clear evidence of the way in which the radicalization of our society is built on many facets of the algorithm, including the tools provided for ad targeting,” said Cameron Hickey, project director for algorithmic transparency at the National Conference on Citizenship.

Questions about the ethics of using data-driven profiling to target political messages are not new. Perhaps most famously, the British political consulting firm Cambridge Analytica purported to create detailed psychological profiles of Facebook users and shared those with the campaign of former president Donald Trump. While profiling has been a part of politics for decades to some extent, figures ranging from former Facebook insiders to Federal Election Commission officials have raised alarms over the kind of microtargeting that social media allows. (The European Commission is also considering including a ban on microtargeting in its landmark Digital Services Act package, which is making its way through the European Parliament and the Council of the European Union.)

That said, The Daily Wire’s targeting choices are widely accepted as routine, its success on Facebook more of a feature of the platform’s workings than a bug in the system, said Katie Joseff, a research fellow at the Center for Media Engagement at the University of Texas at Austin.

“These platforms, when you look at Facebook and YouTube in particular, they want people on there who are engaging their users because then there’s more users and user time overall,” Joseff said. “So [The Daily Wire] is definitely playing into the structure as it was created and doing it well.”

By Corin Faife – This article was originally published on The Markup and was republished under the Creative Commons Attribution-NonCommercial-NoDerivatives license.