You Likely Have No Idea How TV Ratings Work — A Lot More People Are Watching Than You Think

The full explanation about the difference between TV ratings and website traffic


TV is dying, right? I mean, that’s the conventional wisdom now. People are watching Netflix, or videos on YouTube, or Vines.

But here’s the truth. Way more people watch TV than you think. And the problem is the antiquated, incomplete method for counting those viewers, versus the hyper-accurate way website traffic is calculated. Don’t buy the convoluted hype that websites like Buzzfeed or Vice are reaching more people than TV. There’s a reason why they want to get on traditional TV.

The key to understanding this is knowing what Nielsen ratings really tell us, and what they don’t. So to explain, let’s break it down into what ratings are and what they aren’t.


What TV Ratings Are:

Nielsen calculates ratings in two ways — “diaries” and “people meters.” Most are now people meters, which automatically calculate what and when certain households watch. There’s no publicly-released information about how many Nielsen households exist. Back in 2008 when I worked for TVNewser, I was told there were 14,000 Nielsen households at any given time, and of that group, approximately 3,500 are DVR-capable. I’ve heard rumblings that the number has increased by double or triple. So at the highest estimate, let’s say it’s 50,000 households now (with a larger percentage of DVR-capable).

The households are evenly distributed by population throughout the United States, and are roughly meant to be representative of the population as a whole. They also rotate occasionally. But let’s get back to that number. Maybe 50,000 households seems like a lot to you, but to put it in perspective, there are approximately 100 million households with a TV in the United States. That means even at the high estimate of 50,000 households, the number represents .05% of the households. That would be less than 1/10 of 1%.

The ratings that you see — for cable news, for broadcast primetime shows, etc — are then extrapolated out from these Nielsen households. So in this scenario, each household represents 2,000 other households in approximating what the country is watching. The ratings categories that matter are generally “total viewers” (not as important), the 25–54 demo (viewers between ages 25 and 54, important for news) and the 18–49 demo (viewers between ages 18 and 49, important for entertainment and sports). The ratings you see now are also mostly “live + same day” which accounts for 24 hours of the viewer watching on their DVR/Tivo. There are also ratings measurements that account for three-day DVR totals (a version of which advertisers currently buy based on) as well as 7-day DVR totals (what networks would like to become the norm). The demos are what advertisers care about — how the TV networks make most of their money. They are crucial to the financial success of the channels.

Now that there are hundreds of channels, which means there are more options for these Nielsen households, and everyone, what happens if a couple families decide to watch something they don’t normally tune into? Massive swings in the ratings based on an extremely small sample size.

So there’s one issue, but that’s not the most important key to understanding the difference between TV ratings and website “ratings.” Monthly and daily page views and unique visitors are a combined, aggregate number, while TV ratings are an average. To best understand how a TV rating works, let’s say a show starts at 8pmET and has 10 people watching. The next minute, at 8:01pmET, 2 people change the channel to something else, but 4 new people come in and start watching — now there are 12 people watching. So far, there have been a total of 14 people who have consumed any part of this show — 10 at the start, 4 more after the first minute. But the TV rating for this show so far would be 11, since 10 the first minute is averaged with 12 the second minute.

Extrapolate this out over an entire 30-minute or 60-minute program. Maybe this show reaches up to 20 or so viewers at its height, double what it started out, then drops back down toward the end. Maybe the average ends up around 14 or 15. But in total, many, many more people watched.

When you see a rating that calls itself “total viewers,” it most certainly does not actually show you how many total viewers watched that show. To get closer to that number, you’d need to look at “cume,” which is a cumulative rating that isn’t generally publicly recognized but helps to give a better picture of the number of actual viewers Nielsen thinks watched a show.

Just as a single example of how cume ratings affect what you think viewership is, Anderson Cooper 360 at 8pmET averaged 627,000 total viewers in the second quarter of 2015. According to the TV ratings that you see, 627,000 people watched the show each night. The cume rating, though, was 34.2 million people.

So what is more accurate to say is that in the second quarter of 2015, on any given minute of Anderson Cooper 360, an average of 627,000 people were watching. But it reached a TOTAL of 34.2 million people who watched at least one minute of Anderson Cooper 360 during the second quarter of 2015.

And that’s just one hourly program on one channel. But that view gets closer to accuracy when you compare it to monthly unique viewers on websites. Closer — but not there.

What TV Ratings Aren’t:

As you can see so far, Nielsen TV ratings are complicated, fairly incomplete and underestimate how many people are watching. But that’s not it — because what’s perhaps more telling is what TV ratings are not, what they don’t even attempt to be. And that is an accurate representation of which people are watching television that aren’t sitting in their home.

In the past I’ve used something called the Hotel Bar Scenario. Let’s say you’re on a business trip, you head down to the hotel bar to grab some dinner and a cocktail, and find yourself seated between two other patrons. As you start sipping your drink, you glance up to the TVs above the bar, which is playing Fox News’ The O’Reilly Factor.

You take out your laptop while you’re bored and waiting for your meal and open up NYTimes.com. You click on a few articles. The woman to your right has her iPad propped up, and clicks on a Huffington Post link in her Twitter feed, then moves on to her meal and leaves it open. The man to your left has his phone out, and opens up an email from his daughter with a link to a Buzzfeed quiz. He takes it. All of the actions at the bar are registering toward the statistics of these websites, and are giving them lots of valuable data as well. Unique visits, all of your page views as you click through, time spent on page, especially for that Huffington Post link that was just left open and ignored. All of it is positively counting toward the traffic of these websites.

Meanwhile, The O’Reilly Factor plays overhead. You watch periodically while you read the articles and eat your meals. The woman to your right is pretty engrossed in what he’s saying and watching intently. The guy to your left glances occasionally but otherwise is sticking with his burger. But here’s the key — none of your viewing of The O’Reilly Factor counted in the ratings. And not in the same sense as if you were sitting in your living room watching. No, Nielsen isn’t even attempting to count your viewership, through extrapolated data based on households or any other measure. The three of you watched some TV, and it didn’t “count” — it disappeared. As opposed to your consumption of the Internet, which all registered in many important ways.

Think of how many scenarios fit this example — hotels rooms, at work, airports (CNN has an entire airport channel!), browsing at Best Buy. You watch TV a lot, even for a minute or two, without really thinking about it, but more importantly, without your viewership being noted in any way.

How about the Super Bowl. Lots of people watch it, right? This year’s Super Bowl was the most-watched TV program of all time, and the past 5 Super Bowls make up the 5 most watched programs ever. But the rating for this year’s Super Bowl was 114.4 million, meaning, according to Nielsen, 114.4 million people watched. But since we know how Nielsen calculates ratings, what this really tells us is that an average of 114.4 million people watched at any given minute. So it’s not really the total. Also, we know that it only is accounting for people watching in their own households. I watched the Super Bowl this year at a party at someone else’s house. Nielsen isn’t even attempting to count people like me. Did you watch it at a bar or restaurant? You aren’t being counted either. Millions more people consumed some part of the Super Bowl, and are simply not being counted. (Here’s more on the difficulty of counting viewers who watched the Super Bowl.)

For now, there is no way to correct this. Netflix certainly knows things about you — like when you sign into your account, or when and what you’re watching. But traditional TV, still viewed by hundreds of millions of people, decreasing a little and generationally by more, but not enough to be overtaken by Internet-only brands, has no way to truly show how many people are watching. (Nielsen is taking steps to try to address the out-of-home issue, but so far it’s just in the test-phase and somewhat inaccurate. They also are trying to incorporate digital viewing around “TV,” but again, it’s very early in the process.)


So, tomorrow the first GOP debate will be on Fox News. It will get a high number in the ratings. But it will not be anywhere close to telling you how many people actually watched. And it’s important to know that the next time you hear someone tell you about the death of TV.

[Autonomous is a free digital magazine. But you can donate to the mission, and receive a variety of added benefits (including extra content), by visiting our Patreon page.]

Show your support

Clapping shows how much you appreciated Steve Krakauer’s story.