Tuesday, September 14, 2010

Intro to Nielsen Ratings


Last updated: 8/12/2011. As I said when I first wrote this, please let me know if you find anything that's unclear or incomplete or incorrect. I'd like this to lay pretty much all the necessary groundwork you need to start looking at TV ratings, and I could very well have left out something big!

What are Nielsen ratings? They are measurements of how many people watch TV. "Nielsen" is the name of the company that measures audiences for TV and other forms of media.

Why are there TV ratings? To help determine the dollar value of advertising in a given program. Shows are kept around because they are good at bringing in ad dollars, not (directly) because of how popular they are.

How does Nielsen come up with this data? Is it reliable? Nielsen ratings come from tracking the viewing habits of a representative sample of about 25,000 households. The TV-owning United States population consists of 116.9 million households for the 2010-11 season. That means the Nielsen sample is about 0.02% of the population. That doesn't seem like very much, but it is about ten times the size of the sample back in 1977.

What I do know is that while there can be short-term variance in individual data points, most stuff that happens over the long-term makes sense. Most shows follow fairly similar general trajectories. There are a few exceptions, but it's hard to believe there wouldn't be exceptions even if everyone were counted. I also know that pretty much everyone on all sides of the industry seems fairly trusting of the system. That doesn't necessarily legitimize it, but there's a LOT of money being thrown around based on this stuff, so I have to believe there's some general sense among people with a lot more access than I that this stuff is reliable. A recent article by Craig Engler of Syfy said, "All of the other data we look at ... shows people watch on demand, DVD sales (we often get this data even though we don't usually share the profits), digital downloads from iTunes and Amazon, streams on the Internet, visits to show Web sites, even piracy ... give us different metrics to look at alongside TV ratings to make sure nothing really weird is going on."

Most people who have a beef with Nielsen ratings seem to be folks who start with the idea that their individual shows are being undercounted, then leap to "Nielsen is flawed." I'm not aware of any actual numerical case against the validity of the numbers, but if that exists, I'd be quite interested to hear about it.

For more on how Nielsen works, USA's Ted Linhart posted a 20-page booklet from 1977 that explains the ratings system in some detail. Most of the info on the process still applies.

How useful are TV ratings in a world with so many other types of viewing? This may change in the future, but for now the answer is still "very." While records of online streaming, iTunes downloads and other streams are not publicly available, what we do know is that 1) most viewing still happens on the good old fashioned television set and 2) an even greater chunk of the advertising pie comes from TV viewing. The networks' renewal and cancellation decisions still correlate quite closely with shows' abilities to generate TV advertising revenue (which in turn correlate quite closely with TV ratings).

Explain all the numbers on the average TV ratings result. Let me just bring in a sample one from TravisYanan (a must-follow Twitter feed that posts final ratings from every night):
Pretty Little Liars (8pm)
- 3.071 million viewers
- 2.0/4 HH
- 1.1/3 A18-49
- 2.4/8 W18-34
- 4.2/16 T12-17
First line of numbers is the easiest to understand. 3.071 million viewers is just the total number of people who watched the show. Each other line is some form of a rating/share. To the left of each slash is the rating, or the percentage of that entire group in the US that watches the show. To the right is the share, or the percentage of the people watching the show out of a subset of people within that group who are watching TV at that time.

So what are the groups? HH is a rating/share based on the number of US households. Most other groups are based on viewers, not households. A18-49 is adults age 18 to 49, W18-34 is women 18 to 34, and T12-17 is teens 12 to 17. Other key groups to look out for are A25-54 (adults 25 to 54, typically the demo that news organizations go after) and P2+. (persons 2+, or a fancy way of saying "total viewers." Total viewers are usually measured by raw numbers, as above, rather than with a rating/share. Occasionally, though, you will run across a P2+ rating.)

As an example, Pretty Little Liars' 1.1/3 in A18-49 means that 1.1% of all US adults 18-49 viewed Pretty Little Liars either live or within the same day with their DVR, and 3% of all adults 18-49 who were watching TV at the time watched  Pretty Little Liars.

Why does your blog traffic in adults 18-49 and not the total number of viewers? The answer to this question used to be because it seemed like the intelligent people were using and advocating adults 18-49 as the advertiser-preferred demographic. But I've since done a project called Peetooplus (a phonetic spelling of "P2+" AKA total viewers) in which I tried connecting upfront ad rates to total viewer and adults 18-49 averages. I concluded that ad rates correlated resoundingly better with adults 18-49 than with total viewers. Not the greatest methodology in the world, but it convinced me to the point where I often barely even look at the "peetooplus" numbers anymore, and will almost never bring them up here except for when they are interesting for trivia/historical purposes. Adults 18-49 drive advertising and are thus the most correct measurement we have of a show's strength.

I know there's a large contingent out there that thinks the adults 18-49 demo is overvalued, but those people are mostly basing that belief on the way things should be, not the way things are. Maybe persons 50+ are much more valuable to advertisers than the advertisers give them credit for, but I have no way of knowing either way, and frankly it doesn't really matter until someone's able to prove it. What I do have a way of knowing is that the advertisers are currently much, much more interested in adults 18-49. So that's what I go with.

Why did you talk about women 18-34 for some networks? Why don't you talk about them much anymore? I used to believe for those networks, including broadcast net The CW, it's the target audience and the number that drives ad dollars just like adults 18-49 does for the big nets. My first post on "The CW Demo" should illustrate a bit about where I was coming from initially. After the 2009-10 season, when I started regularly archiving W18-34 numbers, I was finally able to compare them with upfront 2010 ad rates, and I found that despite all of the talk about the "CW demo," the network's ad rates still correlated more closely with the same ol' adults 18-49 than with anything else, including women 18-34. There are so few shows on the network that I'm not wholly convinced, and I might revisit it again down the line, but for now I'm mostly back to referring to the network using adults 18-49.

Why does your blog use Live + Same Day DVR numbers when the entire world is watching TV on their DVR days later? Almost all TV ratings discussion here and elsewhere (including the sample numbers above) are "Live + Same Day DVR," meaning they are measurements of how many people watched the show as it was being broadcast plus how many people watched via DVR between the live broadcast and 3:00am of that night. I did a project that dug a little deeper by breaking those ratings into their "Live" and "Same Day DVR" components, but the main reason they're so prevalent is availability. It takes three weeks for Live + 7 numbers to become available as opposed to one day for Live + SD, and even after those three weeks they're often not publicly reliable. I really don't want to wait three weeks to talk about last night's ratings. And that's not even the real data that sets ad rates; those are called C3 ratings, and they measure commercial viewing within three days of the telecast. Those publicly appear even less often. And from what little I've seen of them, Live + Same Day numbers are a pretty good approximation. Most shows' C3 ratings are ultimately within about plus or minus 10 percent of the Live+SD. I'll gladly take a pretty good approximation the next day over the exact numbers three weeks later. (That'd apply even if there were a reliable way of getting the exact numbers, which there isn't.)

All your ratings averages look off. Why don't you just use the ones Nielsen or someone else supplies? Because I don't trust them.  When I talk about an average adults 18-49 rating, what I want to see is a Live + Same Day average of all original episodes and no repeats. Many of the Nielsen-supplied averages are "Most Current" (a combination of Live + 7 ratings when available and Live + SD when the +7's aren't) and many averages that they and other people provide include repeats. So I just hand-calculate them. I might make the occasional human error, but in terms of applying an average to the overnight ratings you see each day, I'll take my numbers over anyone else's. I just want all my numbers to be the same, and if that takes extra work, so be it. Also note that my averages are based on ratings which are rounded off to one decimal point, not exact numbers, so due to rounding I may be a tiny bit off of the actual average, but not by any remotely significant amount.

Why don't you like repeat telecasts in your averages? The main reason is that it isn't a level playing field. Many serialized and reality shows run zero or almost zero in-timeslot repeats. So the shows that repeat well are actually penalized in the with-repeat averages, because the high volume of repeats drags them down when it won't for the other shows. Seems counterintuitive to me, so I stick with originals-only averages.

Calculations. I discussed a couple of these above, but I just wanted to lay them out in another way.

Rating = 100% * number of people/households watching in a group / number of people/households total in the group

Share = 100% * number of people/households watching in a group / number of people/households in that group watching any TV

Total number of people/households watching =  rating * number of people/households total in the group / 100%

The above calculations can be used in tandem with TVByTheNumbers' figures about the TV-owning population to convert audience sizes into ratings or vice versa. (This is helpful because the currency used in press releases tends to vary.)

% Households/People Using TV (HUT/PUT) = 100% * number of households/people using TV / number of total households/people OR 100% * rating / share... Since there is so much rounding in rating and share figures, if you really want to determine the HUT/PUT in a given time period, I recommend doing that rating/share calculation with as many shows as possible and averaging them. It'll probably still be off, but that'll get it as close as possible.

No comments:

Post a Comment

© SpottedRatings.com 2009-2022. All Rights Reserved.