Accurate measuring of online and broadcast viewers is tough biz
It’s probably the single biggest issue vexing the television biz these days: how to measure ratings on the Internet.
As viewing of TV and movies via broadband platforms is becoming more prevalent, there’s a growing need for uniformity and standards among Internet ratings services. At present, it’s a Wild West of dueling providers and methodologies.
“The truth is, there are a lot of measurement companies out there trying to put a yardstick against online streaming,” said Peter Naylor, NBC Universal’s exec veep of digital ad sales. “All of them report different numbers, because they’re all panels made up of different people.”
In TV, there’s always been just one currency: Nielsen ratings. The higher a show’s rating, the higher the ad price for a spot on that show. But for the online video market, it’s much harder to answer key advertiser questions: “How many people are watching this show? How old are they? And how much money do they make?”
Rob Tuck, head of advertising sales for CW, says the question of how to leverage online ad sales is “the biggest challenge that the industry has been facing.” For all the talk of booming digital advertising dollars and the death of the old-fashioned 30-second TV spot, buyers eager to drop the same dollar amount on webcasts as they did on broadcasts remain few and far between.
“Most of them wanted to get involved,” Tuck said, “but we kept hearing that the challenge was ‘I don’t know how to bring (the Web) into my national buy when I need the same demographic guarantees I get with TV.’ ”
The big focus for traditional networks is developing a means of combining the ratings for traditional live TV viewing with DVR and online watching into one cume number that could be sold to Madison Avenue. Getting there is a lot harder than it sounds. A clutch of major showbiz congloms and advertising agencies banded together in fall 2009 to create the Coalition for Innovative Media Measurement, designed to fund research into building a better yardstick. (It was also intended as a spur in Nielsen’s side to step up its development of integrated TV and Internet ratings.)
On April 25 Nielsen will roll out its Extended Screen service to track TV and Internet viewing, with ratings data retroactive to mid-March. But it will be able to integrate TV and Internet data only if the program carries the same advertising spots in both platforms.
That has not been the case for most TV programming offered on the Web by the major networks — the ad load online has typically been at most one-third the number of spots that would run in a traditional telecast — but the nets are moving in the direction of synching up their ads in the hopes of wringing more coin out of Internet exposure for their programming.
Learning the difference
Another challenge for the traditional TV biz is the lack of a central source for online viewing stats. As Internet video has taken off thanks to Hulu, Netflix and the like, the major purveyors of Internet ratings are Nielsen, DoubleClick (which provides vital user-counting products) and Reston, Va.-based ComScore.
To tackle the problem, ComScore and Nielsen had to reimagine the way ratings work. Nielsen takes its television statistics from a panel of viewers assembled by the company to represent a microcosm of the U.S. population. How well those panels actually reflect the demographic complexity of the nation’s 110 million-plus television households is often a point of contention with the broadcast and cable networks.
In many ways, the new world of online viewing allows for far greater specificity in tracking auds. Through third-party companies like DoubleClick, networks with a big online presence can demonstrate to a client that they’ve served a particular ad an exact number of times, because the servers hosting the ads keep track of how many times they’ve been played. Media publishers have access to their own data, of course, but they need a third-party system for tallying it up in order to stay credible, which is where DoubleClick comes in. The company provides what it bills as “detailed and actionable reporting” on just how many people have viewed ads this week (or month, or quarter). And, unlike the ballpark figures provided by Nielsen TV ratings, these numbers are exact. At least they’re supposed to be.
Showing off the number of ad plays (“impressions”) isn’t the same as proving that you’ve reached a certain number of people, though. Browsers like Firefox can be tweaked to reject cookies (the data packets that tell a website you’ve visited it before); also, users might watch from multiple locations or devices, and they might clear their browser caches, making them look like a new unique user to DoubleClick’s servers. The number of “unique” users can get absurdly large.
“We’ve seen cases where some large portal-type publishers (like television networks) will have more cookies than there are people in the United States,” said Andrew Lipsman, ComScore’s senior director of industry analysis.
Two kinds of measurement
Tuck and other ad sales professionals say they need the online equivalent of Nielsen families in addition to their own server data in order to demonstrate an ad’s efficacy. With the right data, you can say not merely that you’ve served an ad 5,000 times, but that 25% of the people who watched that ad were in, say, the 18-34 demo. You need DoubleClick, but you also need another source.
“We take two sources,” Tuck said. “DoubleClick basically provides the equivalent of (total viewers), and then we took the metrics from Nielsen census data, and we applied the audience demographic to that (total viewer) impression … and that provides a demographic impression. The clients are now able to see what the online buy delivered on their demographic.”
Creating the perfect panel …
That demographic info is where the competition gets serious. Nielsen, ComScore and several other hopeful contenders have put together comprehensive panels of users who represent desirable consumers as accurately as possible, each claiming superiority but none quite achieving total market control. ComScore’s sample encompasses 1 million people; Nielsen’s is 250,000.
In the struggle for dominance between Nielsen and ComScore, the goal is to create a panel that is both an accurate representation of the viewer landscape and the largest possible cross-section of users you can find.
The proposition is “challenging and expensive,” said Matt O’Grady, Nielsen’s exec veep of media product leadership. Lipsman and O’Grady agree that viewers in certain demos are tough to come by: men 18-24; young, nonwhite men; affluent viewers of any age — all present unique challenges of one kind or another, even to an organization with an unlimited budget.
Lipsman said his company has seen positive results by offering a wide range of incentives to hard-to-nail-down panelists. One wealthy Web user might not care about a new videogame or a gift certificate to Amazon.com, but a tree planted in a Third World country for every month she stays on the panel might create the necessary interest to get valuable raw data from her.
… while avoiding some major problems
There are pitfalls, of course. “If we don’t offer a wide enough range (in the sample), that creates selection bias,” Lipsman said.
Then there are less-scrupulous competitors. Other orgs provide much larger groups of consumers, said one exec, but bigger isn’t always better.
“At the end of the day it comes down to how well you take precautions to make sure you’re getting as representative a section as possible,” said an insider. Some companies try to pad their data with information from Internet service providers, but that date tends to reflect older, less affluent users who may be using a dial-up connection (the Internet equivalent of a Model-T car) and less likely to shop online.
There are also big-time privacy concerns around ISPs giving out ratings and demographic information.
Yet another difficulty is that quite a bit of online video watching goes on in the workplace. Not many employers are keen on seeing their terminals co-opted by a data-mining company, and even fewer relish the prospect of a work force more interested in catching last night’s “30 Rock” than doing the work they’re paid to do.
More competitors, better numbers
NBC U’s Naylor and others see an upside in the competition among research firms to become the gold standard of Internet ratings.
“We look to ComScore to understand our audience,” Naylor said, noting that Nielsen is getting attention “because of who they are” and that competitor Quantcast also has “an impressively large panel.” The many data options, he said, give media publishers a more accurate idea of what their viewership actually thinks of their product.
Lipsman is quick to point out that multiple sources mean a more-accurate understanding of the demo landscape, especially when compared to, say, a medium where one company measures for the entire industry. But he also understands the reluctance to adopt multiple sources of information.
“The more data you have, the more complex it gets,” he said. “If you have a single number, even if you don’t like it or you think it’s undercounting you, it’s much easier to understand or interpret.”
Execs who’ve spent years marketing a single product based on a single data source now have their work cut out for them. Does the client want a simulcast with the exact same ads he bought for the broadcast version of the show? Does she want to bypass broadcast altogether? Does the company you’re dealing with today prefer Nielsen, ComScore or Quantcast numbers?
But there’s also been plenty of grousing in the TV world over the past several years about the biases inherent in the Nielsen TV rating process. Nielsen isn’t sitting idle, and its programs such as Extended Screen demonstrate just how fast the video content marketplace is evolving.
“Very few people have the stomach to stay in the business, but competition forces everyone to stay on their toes,” Nielsen’s O’Grady said. “Measurement is good for the collective pond, so it’s good to have multiple measurement sources out there. The TV world is very complex.”