How to use Facebook’s mostly-meaningless reach stats and still do some factual ratings reporting


I’m picking on Axios because the first sentence of its Mission Statement is:

We are a new media company delivering vital, trustworthy news and analysis in the most efficient, illuminating and shareable ways possible. “

At least in one instance they aren’t coming close.

In its reporting of Facebook’s streaming statistics versus the TV numbers, Axios wrote:

Parts of fired FBI Director James Comey’s Thursday Senate testimony were seen by 89 million people on Facebook. Nielsen estimates 19.5 million people saw it on live television.

Why this matters:The high-profile Congressional hearing (McCarthy, Iran-Contra, Anita-Hill) used to be prime spots for television. Now those audiences are shifting online. In addition to Facebook, 2.7 million people watched it on Twitter’s Bloomberg co-branded livestream. (One caveat: a Facebook view can be as short as 3 seconds, so a direct comparison to TV ratings is inexact.)


1. the author doesn’t understand what the TV numbers mean or is being deliberately deceptive (I’m guessing the former, and that…)

2. the author doesn’t realize that they don’t understand what the TV numbers mean

3. Facebook’s reach stat (“watched at least 3 seconds!”) is compared to TV’s average minute audience. As far as ratings reporting goes, that kind of wacky comparison should be considered an abomination.

The average minute audience is the total minutes of viewing for the telecast divided by the duration of the telecast. It represents the average minute at any moment of the telecast. Actual viewership who watched any part would’ve been much higher.

Meanwhile, had Facebook reported its own average minute audience it would’ve been much, much smaller than 89 million.

If Axios is striving to get the information out and be trustworthy, they could’ve still done so, even if Facebook stonewalled them on additional data.

It requires understanding what the numbers mean and being clear in what is being reported. Something like this would’ve worked:

Parts of fired FBI Director James Comey’s Thursday Senate testimony were seen by 89 million people on Facebook. The 89 million is a total reach statistic. It is the total GLOBAL audience of people who watched at least 3 seconds of the telecast. Sadly, Facebook did not report an average minute audience, if they had we project it would’ve been in a range of “less than 1 million viewers to as high as 5 million viewers.

Nielsen estimates that at any point during the Comey testimony, on average 19.5 million people were watching across the six biggest networks televising the testimony or at least four times as many people as the high end of our Facebook projection, despite the fact that Facebook’s statistic is worldwide and the Nielsen TV average minute audience is U.S. only.

Why it matters blah blah blah

Sure, that approach doesn’t necessarily line up with the narrative the author was going for here, but unburdened with any journalism background I just kind of assumed that a site like Axios that is striving to deliver vital and trustworthy NEWS would adjust the narrative to the facts. Multiple people have informed Axios the above story is just plain wrong, and…it’s still up.


Another benefit of the above approach: it might help get Facebook in the mode of reporting total minutes viewed and/or average minute audience, especially if your internal estimate is lower than the actual number!

This article has 1 Comment

  1. I have lots of stuff on in the background especially on twitter….I rarely watch for more than 5 seconds and rarely turn on the sound. So any viewing ought to include – maximise the window and turn up the volume…and actually watch. Being British a lot of the American twitter videos eg Lacrosse, Bloomberg may be on but are not watched.

Leave a Reply