Trump's inauguration attracted more than Nielsen is saying

Streaming counts total viewers, while television counts the average per minute

Editor's note: Our Splash Health, Wellness and Wearables event is coming up on March 23 in San Francisco. We'll have Mario Schlosser, Founder & CEO of Oscar Health, Brian Singerman (Partner, Founders Fund), Steve Jurvetson (Draper Fisher Jurvetson), J. Craig Venter (Human Longevity), Lynne Chou (Partner, Kleiner Perkins), Michael Dixon (Sequoia Capital), Patrick Chung (Xfund), Check out the full lineup and register for tickets before they jump! If you’re a healthcare startup and you’re interested in being part of our competition, learn more and register here.

Also, vote for your favorite healthcare startup before February 16! Vote here!

There's no understating the importance of the shift over just the last few years in how we watch television. With the rise of cord-cutters, and so many channels, as well as cable companies, offering a streaming-only option, it may seem as though what we think of as traditional television is going the way of the dodo.

While there's certainly the distinct possibility that streaming numbers have begun to pass television views, it's actually hard to know if that's true or not right now, simply because the two aren't counted the same way. 

To fully understand what I mean, let's look at Donald Trump's presidential inauguration this past Friday.

According to data put out by Nielsen, a total of 31 million people watched the event on average across 12 channels, including ABC, CBS, NBC, TELEMUNDO, UNIVISION, CNBC, CNN, Fox Business Network, Fox NC, Galavision, HLN and MSNBC. That makes it the fifth most watched inauguration, behind only Ronald Reagan in 1981, with 41.8 million; Barack Obama in 2009, with 37.8 million; Jimmy Carter in 1979, with 34.1 million; and Richard Nixon in 1973, with 32.9 million. That means Trump attracted a pretty high number of television viewers.

The question today is, however, what happens when you add in streaming viewers? After all, we all know that more and more people are watching online. 

In fact, if we add in all those viewers, Trump's inauguration viewership (which has stirred up way too much controversy) may have actually garnered significantly more views than Nielsen is counting.

To wit, CNN had 7.2 million unique live stream viewers of the inauguration. On Twitter, meanwhile, there were 6.8 million unique viewers who live streamed the event on PBS Newshour, slightly more than those who watched on Election Night, making it the largest Twitter live streaming audience ever. Notice how PBS Newshour wasn't included in the Nielsen numbers. That means there was an additional 6.8 million who viewed Trump's inauguration on top of the 31 million. 

Together, CNN and Twitter had 14 million viewers, compared to 31 million viewers on TV. And that's not counting other platforms like YouTube and Facebook, neither of which have yet released any data on their audience size.

When I reached out to Facebook, the company told me that there aren't any current plans to release the number of streamers only numbers around likes, posts and comments.

"During Inauguration Day, January 20, 60 million people on Facebook in the U.S. generated 208 million likes, posts, comments and shares related to the presidential inauguration," a company spokesperson told me.

I also reached out to YouTube, but the company was unavailable for comment at this time.

However, if those other platforms had 17 million combined streamers then it would equal the number of viewers who watched on a television.

Calculating TV ratings vs digital views 

Except, not.

While it seems like streaming likely did outdo live TV, there's a problem: the way streaming is counted is not the same as television, and it skews the numbers.

For streaming videos, everyone who watched, even just a second or two of the video, is counted. That means that if someone accidentally clicked on the page, and then clicked back, they would still be counted as a viewer, even if they didn't actually watch the video or even know it was on the page. 

TV ratings, on the other hand, count the average per-minute audience. That will be significantly lower than the total viewing audience spread out across an eight-hour long program. 

As Steve Hasker, Global President and Chief Operating Officer at Nielsen, explained in a blog post in 2015, the way these two metrics are presented "unfairly tilts the comparison against TV. "

He gave two examples of how the numbers can change depending on which metric you are looking at.

First, he examined the numbers for Jimmy Kimmel Live in May 2015. On YouTube, the audience was 9 million on average for each episode, with some gaining over 25 million views, while the average audience was on television was only 2.2 million. Going by those numbers alone, it seems like streaming significantly beat TV in the number of viewers.

The actual numbers are much closer than that. Kimmel's audience on television was actually 5.3 million on average for each episode. The 2.2 million number is just how many people were watching in a given minute. YouTube may still have had more views on average than TV, but using the average-per minute metric actually makes the gulf look wider than it is, and puts TV at a disadvantage. 

Haskel also looked at the 2014 World Cup on ESPN, which had an average-minute TV audience of 4.6 million persons, and had 115.5 million digital views. Again, the numbers are skewed, he said: the average-minute digital audience was just 307,000, much lower than the initial numbers suggested. 

To understand how much those one two-second videos can throw off the numbers, just look at what happened when Facebook was recently caught falsifying their own numbers by excluding them. 

The company had been telling advertisers that it was calculating the Average Duration of Video Viewed by dividing the total time that the video was watched by the number of people who played it. It turned out, though, that the company wasn't counting total numbers, only those who played the video for at least three seconds. By omitting those shorter views, it was artificially inflating the Average Duration. 

This caused enough of a dicrepency that The Wall Street Journal reported that Facebook told ad buying agency Publicis Media that it has overestimated average time spent watching videos by between 60 percent and 80 percent. 

What that really means is that, with those one or two second long views counted in an average per-minute audience metric, they would likely significantly lower the numbers for streaming services. 

Given the way the numbers are counted, the total number of viewers on television for the inauguration was very likely much higher than those on streaming. That's just never how they've been measured, and it probably wouldn't be fair to start now. 

Also complicating these numbers is the fact that there may be overlap between the two. On CNN, for example, there is really no way for anyone to know if people are on multiple devices at the same time, so the number of users on streaming may not even be taking away viewers from the Nielsen ratings.

Ultimately, the different metrics make it hard to know how well streaming is doing compared to live television, and trying to compare them just isn't very fair. 

"Both TV networks and digital-only video publishers need to use fair comparisons when talking to advertisers. Underlying this, we need independent third-party measurement for non-traditional video channels that is comparable to what we provide for TV," Hasker wrote as a solution to these problems. 

Until then, it won't be possible to know if, or more likely when, streaming actually does overtake television ratings. 

(Image source: corporate.comcast.com)