Facebook argues its own tool showing conservative dominance isn’t ‘fully representative’

Listen to Facebook, just not a Facebook-owned analytics tool designed to provide insight into Facebook. 

That appears to be the argument made late Tuesday by Facebook’s Alex Schultz in a frankly confounding blog post titled “What Do People Actually See on Facebook in the US?” Shultz serves as both Facebook’s vice president of analytics and the social media company’s chief marketing officer, a combination that perhaps helps explain his bizarre insistence that the Facebook-owned CrowdTangle tool, which consistently shows right-wing links dominating the platform, isn’t “fully representative.”

CrowdTangle, an analytics tool Facebook makes available to “select Facebook publishing partners,” is designed to help users “[benchmark] the performance of [their] social media accounts vs those of [their] peers.” Using the tool, for example, you can see which public posts had the most interaction during a certain time period. 

Schultz would have you believe that, if you truly want to understand what’s happening on Facebook, you should pay no attention to these Facebook-generated data points. Instead, please pay attention to the different set of Facebook-generated data provided in Tuesday’s blog post.

CrowdTangle’s version of reality

Over and over again, CrowdTangle shows conservative pages dominating Facebook. 

For example, according to CrowdTangle, as of the time of this writing the 10 posts with the most engagement over the previous 24 hours were from: Donald Trump, Donald Trump, Cristiano Ronaldo, Barack Obama, BTS, Donald Trump, Donald Trump, Laura Clery, Leo Messi, and Donald Trump. 

Earlier today, around 11:30 a.m. PT, the top 10 posts with the most engagement over the preceding 24 hours were from: Donald Trump, Donald Trump, Donald Trump, Donald Trump, Donald Trump, Cristiano Ronaldo, Donald Trump, BTS, Dr. Ben and Candy Carson, and Joe Biden. 

A popular Twitter account, Facebook’s Top 10, run by New York Times reporter Kevin Roose, uses CrowdTangle data to “post the sources of the 10 top-performing link posts by U.S. Facebook pages every day, ranked by total interactions.”

This Twitter account, using CrowdTangle data, makes clear that there’s a pattern in which type of Facebook pages lead the driving of clicks. 

But, as Schultz writes in Tuesday’s blog post, “[ranking] top Page posts by reactions, comments, etc. doesn’t paint a full picture of what people actually see on Facebook.” The “Most interactions” metric, he argues, doesn’t account for “reach” — that is, “how many people actually see the content[.]”

A chief marketing officer’s version of reality

Schultz, and by extension Facebook, would have you believe that its own analytics tool doesn’t do a good job revealing what’s truly happening on Facebook. And maybe that’s true. Or, perhaps, the company just wants you to look at a different dataset. 

Or, more to the point, perhaps Facebook wants you to look at a very specific dataset of its own choosing that, by extension, paints a picture of its choosing. In fact, Schultz does indeed present his own version of reality apparently designed to counter CrowdTangle’s data.

“Here are lists of the US news publishers and the US-based Pages (based on the domain in their URL) who reached the most people in the US in the week following the last presidential debate,” reads the blog post. 

Indeed, these neatly tailored lists — which look to have been custom built to prove Schultz’s point — do appear less Dan Bongino heavy. Neither that lack of Bongino, however, nor Schultz’s assurances prove that his version of reality is closer to the actual truth. 

That would take full access to Facebook’s data — something the company understandably will not allow. In the meantime, researchers, journalists, and concerned citizens will simply have to use the next best thing to analyze Facebook: CrowdTangle.