Last May, Nielsen Holdings made a shocking admission. The enterprise whose name and famous Nielsen ratings are synonymous with tracking what Americans watch on TV confessed that it had measured viewing time incorrectly, due to technical glitches—not just briefly, but for about a year. It was a failure that cost its most important clients, the major broadcast networks, billions in lost advertising revenue, according to industry estimates that later emerged .
To generate its industry-dominating ratings, Nielsen relies on “panels,” households that click a special remote to show who in the family is watching TV, and what and when they’re watching. In its mea culpa, Nielsen said that 9,400 of its 41,000 panel homes—almost a quarter of the total—had misreported viewing numbers in the period from February 2020 to February 2021. Some sent faulty signals that undercounted audiences, and others sent no signal at all.
The culprit, Nielsen said, was the pandemic. Normally, all those homes would get regular on-site maintenance. But the company had postponed countless repair visits, it explained, because it felt obligated to follow CDC recommendations and avoid sending workers into situations where they could expose customers to COVID-19 and vice versa. Nielsen had fixed what it could remotely, but the problems had dragged on.
The networks that depend on Nielsen’s numbers to determine what they charge advertisers—paying handsomely for the privilege—were livid. As early as the summer of 2020, they had been complaining to Nielsen CEO David Kenny that the sharp drop in their viewer numbers couldn’t be right, especially at a time when families in quarantine were huddling around their TVs for psychic warmth. Kenny had insisted to incredulous network execs that the panels’ ratings were as reliable as ever. Now he was admitting the complainers had been right all along.
Nielsen has neither confirmed nor contested the industry’s estimates of lost revenue. But there was more than just money at stake. Even as he wrangled with the networks, Kenny was aiming to persuade them to stay with Nielsen as their yardstick for audience measurement as they made the wrenching but lucrative transition to a streaming-ad model. Now Kenny’s challenges had multiplied: He would have to simultaneously orchestrate a major technological transition at his company, while mending fences with angry customers.
At the networks, irritation bubbled to a boil. On an August earnings call, CEO David Zaslav of Discovery, which uses Nielsen as its primary scorekeeper for cable and broadcast, suggested the company might not keep that place for long. “Nielsen can’t get its act together, and the answer is that we’ve lost money,” stated Zaslav. “I don’t have a lot of hope for Nielsen. From a technology perspective, [the industry] will leave them in the dust.”
The Video Advertising Bureau (VAB), an association of networks and cable companies that monitors TV viewer measurements, publicly accused Nielsen of systematically shortchanging networks’ audiences. It lobbied the Media Ratings Council, a nonprofit group that sets ratings standards, to strip Nielsen’s accreditation for cable and broadcast TV, which the MRC did on Sept. 1.
“Nielsen did three things: deny, deflect, and delay,” VAB chief Sean Cunningham told Coins2Day recently. “And the denials were coming out of Kenny’s office.”
All told, Nielsen finally acknowledged, its viewership reports had lagged actual totals by 1% to 6%. “In that period alone, advertisers paid the networks $1.6 billion to $2.3 billion less than if the audience had been counted correctly,” says Cunningham. “That’s a loss of as much as 5% of total ad revenues.”
In December, Nielsen delivered another jolt to its clients: It disclosed that since September of 2020, a shift in its measurement methodology caused it to undercount “out of home” viewership—panelists who watch television in, say, sports bars or restaurants using its “portable people meters.” The out-of-home undercount, by the VAB’s estimates, erased an additional $700 million in revenue.
Nielsen has neither confirmed nor contested the VAB’s figures for lost revenue. But Kenny, an advertising- and tech-industry veteran who took the helm at Nielsen in December 2018, admits today that he made big mistakes. “I fault myself for not communicating better with our clients on what could go wrong,” he tells Coins2Day. “We should have been more humble.” Nielsen had restored all in-home service by March 2021; it’s now lobbying to regain its MRC accreditation.
Still, Kenny insists that the broadcasters’ frustration with Nielsen has ulterior motives. It’s chiefly a campaign, he says, to mask how fast digital and streaming are stealing their audience. “We’re the referee, and the referee isn’t always loved,” Kenny says. “Put another way, we’re just delivering the news, and for the networks, the news right now isn’t good.”
That migration from broadcast to streaming may indeed turn into bad news for the networks. But Kenny aims to make sure that it’s good for Nielsen. Indeed, he hopes to make the company just as dominant in measuring audiences for streaming as it is in broadcast, with leading-edge hardware and data-crunching technology—and a surprisingly prominent role for the much-criticized, old-fashioned panels.
As the digital barbarians storm the gates of broadcast TV, Nielsen hopes to seamlessly join, and eventually lead, the audience-measurement revolution. That goal presents both technical challenges and major business hurdles. But Nielsen and Kenny are dealing from a position of strength: Despite the anger generated by Pandemic PanelGate, the number of major TV customers that have withdrawn significant business from Nielsen since 2020 is…exactly zero.
What Alphabet’s Google is to internet search or Meta Platform’s Facebook is to social media, Nielsen is to TV ratings. Dominant since the dawn of television, to this day this 99-year-old stalwart holds a near monopoly over the audience measurements that determine which sitcoms or police dramas get canceled or extended, and what advertisers pay for airtime. The Nielsen panels have long been virtually the only option for knowing which programs reach, say, affluent single men ages 25 to 35, the demo most likely to buy a new Toyota or Chevy.
Nielsen isn’t huge by the standards of the broadcast giants it measures; its annual revenues are about $3.5 billion. But its metrics set the prices for 95% of the broadcast and cable airtime sold in America, guiding some $65 billion in annual spending on commercials. Nielsen also works for advertisers by polling viewers about what they bought after watching their ads, a unit called “audience impact.” The leading broadcast and cable networks reportedly each pay Nielsen over $100 million a year, and its commanding role helps it garner sumptuous profit margins of 42%.
But these days, that profitable kingdom is under siege. Even before the undercounting scandal, some of Nielsen’s customers had begun claiming that their long-standing partner is too antiquated to serve as tomorrow’s gold standard—especially as more audiences move to streaming. For the first time in years, they’re looking to replace their legacy provider with a group of hungry, tech-heavy invaders.
That’s why Kenny argues that he’s the guy to preserve Nielsen’s dominance. Kenny spent his career upending old regimes, first as founder of pioneering digital ad agency Digitas, then building the online business at ad giant Publicis, and even converting the Weather Company from a fading TV channel to a hit mobile app. “He’s all about tilting at what look like windmills, then you find he’s conquered the castle,” says Dave Morgan, chief of TV ad buying platform Simulmedia, who was a supplier to Kenny when the latter worked at Publicis.
Kenny has already gone a long way toward transforming a bloated non-innovator into a big data and A.I. Juggernaut. “I had to disrupt Nielsen to disrupt the industry,” he says. Like many others, Kenny is convinced that in a few years, the vast bulk of viewership will move from cable and over-the-air broadcast—what’s called “linear TV”—to the Netflixes and Amazon Primes and notably the ad-supported Hulus and YouTubes that will devour more and more commercial time and dollars.
By its own estimates, Nielsen is already the leader in measuring digital audiences, guiding around half of the advertising sales tied to streaming platforms. But that market is still modestly sized, and many streamers use their own yardsticks. Kenny aims to deploy his expertise in crunching data and computer modeling to make Nielsen the leader in the new marketplace.
At the same time, broadcasters are moving to establish their own ad-sponsored digital services. Disney’s Hulu, Fox’s Tubi, and Google’s YouTube are already using Nielsen. If Nielsen can establish itself as the “currency” for counting digital viewership with big streamers, it’s also likely to get the networks’ digital business. And since cable and broadcast will remain a big part of the market for a while, the holy grail for measurement will be applying the same metrics to digital and linear TV—allowing advertisers to compare prices and audiences across both mediums.
That product doesn’t exist yet in any widely accepted form. But it’s needed for a viewership that’s quickly shifting from the prime time “appointment” paradigm to one of watching anytime you want, on any device. Viewers “will only care about scheduled TV when it’s a live event like the Super Bowl or the Oscars,” says Kenny. “We’re going to help change the landscape from a programmer- to an audience-led market.”
Kenny’s strategy is to develop new metrics for digital viewers and to convince customers to use those metrics for all audiences. The streamers are giving his blueprint good reviews, and if Kenny succeeds, the networks will likely have to accept the Nielsen system. But if the strategy falters, Nielsen’s business could collapse as customers splinter to such rivals as Comscore, iSpot.tv, and VideoAmp.
In a sense, Nielsen and the traditional networks now face the same dilemma in a time of transition: “Being big and the market leader doesn’t guarantee success,” Kenny says. “Only being the innovation leader does.”
Here’s the challenge and the opportunity: As viewers leave traditional TV for streaming, they’re greatly reducing the audience that’s watching advertising. But the streaming audience could eventually be far more valuable to advertisers than today’s broadcast-and-cable viewers are.
Since 2016, the old-line TV audience, comprising households watching cable, has shrunk from 100 million to 72 million , and it’s forecast to fall to 57 million in 2026, according to research firm Insider Intelligence (formerly eMarketer). In the same period, the number of homes whose TVs connect to streaming services has jumped from 78 million to 107 million, on the way to a projected 117 million in four years.
The headline result of this tectonic shift: According to Nielsen, as of December streaming accounted for 28% of all viewing minutes, while broadcast and cable harbored 63%—down from an estimated 95% just three years ago.
Most of the big streamers, including Netflix, Amazon Prime, and Disney+, don’t air ads at all. And a smaller group that does show commercials, such as YouTube, Disney-controlled Hulu, NBC’s Peacock, Amazon’s IMDb TV, and Roku Channel, find that their economics work best when showing fewer ads than run on traditional TV. As a result of all those factors, only 5% of all TV advertising time appears on streaming services, despite their almost 28% share of viewing time. And 95% of all ad minutes still populate traditional TV, even though about one-third fewer people are watching those spots.
Despite the decline in viewership, big advertisers aren’t cutting their commercial budgets. Traditional TV is still a great place for feel-good, expensively produced spots that build brand loyalty for truck-makers and soda-sellers. “Cable and broadcast still have the advantage of making a few buys, and getting a very large audience in one place, where you’d need hundreds or thousands of buys in digital,” says former Time Warner executive and Akamai CEO Paul Sagan. Since fewer people are watching, carmakers, insurers, and packaged goods purveyors are competing to buy more advertising time to reach the same number of viewers as a few years ago. Put another way, ad rates per “eyeball” are rising.
The upshot is that total cable and TV ad spending is declining only slightly. In 2021, the networks and cable broadcasters collected $65 billion by selling ads, down around $5 billion from 2017. By contrast, outlays on streaming ads jumped from $2.5 billion in 2017 to $14.4 billion in 2021, on their way to a projected $34.5 billion in 2025.
Streaming viewers are typically younger, higher-income folks than the cable and broadcast crowd—a very attractive demographic. That’s one reason why, for each viewer reached, advertisers are paying a lot more on streaming services. That in turn explains why the internet platforms are already getting a big share of revenues on a puny proportion of overall ads. Kenny believes those trends will continue: Even as streaming expands, the video universe will continue to offer fewer advertising seconds than in the past. “That shrinkage means each ad second will be more expensive, making the measurement metrics even more important,” he says.
What makes the streaming audience even more attractive is the way it enables companies to show ads to narrowly targeted groups. In linear TV, everyone watching the same program sees the same ad most of time. The networks are working toward greater targeting. But for now, digital ads have more advanced technology and data for zapping spots specifically to folks that advertisers predict will be especially interested in their car insurance or pain reliever. And advertisers are going to want somebody—whether it’s Nielsen or a competitor—who can reliably tell them which streaming services can best help them reach those targeted groups.
But for now, the fast-growing demand for ad time on streaming is far outstripping the tight supply. “Marketers looking to ad-supported streaming as a golden goose to get back the people they lost on cable TV aren’t going to find the inventory they need to offset the loss of linear TV,” agrees Bharad Ramesh, executive director of research and analytics at GroupM, a division of ad giant WPP and one of the world’s largest purchasers of TV space for the likes of Unilever and Coca-Cola. “Marketers need to redefine premium video to include social media and in-game advertising––two vibrant and growing opportunities.”
Kenny has already made Nielsen a pioneer in capturing the share of time spent across all types of TV. Last summer, the company introduced a measure called the Gauge that shows the portion of minutes spent by viewers on local broadcast, cable, streaming, and the “other” category (which consists mainly of video-game viewing). For Kenny, it’s crucial to also measure audience numbers in the non-ad, subscription services that currently dominate streaming. In fact, he says, some of the subscription-only streamers are already paying Nielsen to count their audiences, though he declines to say which ones.
Netflix does its own surveys, but the company says it likes the Gauge and Nielsen’s approach to getting the full picture of where America is spending its time. In an interview with the New York Times in June, Netflix co-CEO Reed Hastings stated that "Nielsen’s in a good place to referee or score-keep how streaming is changing the U.S. Television landscape.” Hastings added that Nielsen “used to not be very sophisticated or accurate view of in terms of internet viewing,” but their technology has markedly improved. “They now have an accurate view,” he concluded.
A case in point is a recent agreement between Nielsen and Roku. In Q4 of 2021, the Roku Channel reached U.S. Households with a total population of 80 million, offering a vast variety of programming, ranging from cooking and travel shows to old movies to such current fare as Showtime’s Billions. For years, Nielsen has been helping Roku deliver ads on the Roku Channel so that marketers can buy from Roku just as they would on traditional TV.
Under Roku's new Nielsen alliance, Roku’s ad buying platform One View will apply Nielsen’s measurement to all four screens: traditional pay TV, streaming, desktop, and mobile. It’s all about revolution in cross-screen measurement. “The more advertisers know about the full audience we’re reaching, the more dollars they spend with us,” says Louqman Parampath, Roku’s VP of product management. “The transition to advertising over streaming will get faster and faster.”
Until recently, Nielsen was measuring what more than 300 million Americans watch on TV by polling just 41,000 households, via its panels. But the splintering of viewers among scores of streaming competitors overwhelmed that system’s ability to track viewership. It’s as if a political poll that accurately surveyed candidates in a two-party race suddenly had to poll for 50 contenders. “The sample size of the many programs and time spots our households could measure got too small,” says Karthik Rao, Nielsen’s chief operating officer.
Kenny believes that the best solution to this conundrum, and one that enables advertisers to reach the ideal total audience, is to combine the scope of big data with the up-close-and-personal, household-by-household insights garnered from its panels. It’s big data, augmented by A.I., that feeds the Gauge, Nielsen’s new yardstick.
Nielsen harvests data from three main sources. The first is viewer numbers from cable and satellite TV services. Cable operators such as Comcast and Charter, and satellite services DirecTV and Dish, collect “return path” data that shows what channel each household is watching, and at what time. The second fount of information comes from smart TVs: Hardware manufacturers such as Vizio, LG, and Samsung, as well as Roku TVs featuring its operating system, apply a technology called automated content recognition (ACR) to determine what shows and ads each of their households are viewing. The third source is viewing numbers from streaming app owners such as Roku, Tubi, and Alphabet’s YouTube.
All told, Nielsen filters data from 100 million set-top boxes and smart TVs. But those gigantic feeds alone don’t create the vivid audience profiles that advertisers prize. To fill in the gaps, Nielsen builds computer models that deploy A.I. And machine learning. It learns where viewers live by purchasing address information from the cable providers and streaming services. Census surveys provide data on family sizes and incomes across 42,000 zip codes that cover those addresses. Nielsen also buys marketing information from sources that range from magazine publishers to health care providers, to help estimate how much of its audience data set includes, say, newly married women or elderly homeowners. All that data, run through sophisticated algorithms, helps Nielsen advise advertisers on how best to reach the audiences they seek.
The number-crunching doesn’t stop there. Nielsen can obtain data on only about one-third of the total cable and streaming audience, because many providers don’t share the relevant information. The data also often excludes, say, additional TVs in a multiple-set household. So Nielsen uses further modeling to extrapolate from the data it does have to cover all 121 million households with TVs—estimating who’s missing from its huge database, and what those absent households are watching.
The models serve two additional roles. First, machine learning learns from new data to accurately predict changes in viewing habits. One recent trend that Nielsen’s models spotted: Families that watch live sports in the evening now stream Hulu or YouTube a lot more in the afternoons before switching to the games on cable. Nielsen’s models can predict how fast that trend will accelerate from year to year, guiding advertisers to buy streaming time before big events.
Second, big data doesn’t tell you who’s watching in a multigenerational home––something the panels can report so precisely. If it’s mostly the kids, not dad, who are watching, raw numbers can send the wrong signals. But A.I. And machine learning can provide strong probabilities on whether it’s grandpa, the teenagers, or the parents watching given shows at given times, based on the mix of household metrics.
For Kenny, big data is now replacing the panels as Nielsen’s main act. But he still believes there’s a crucial role for those old-fashioned audience meters: They validate the results from the huge feeds. “Models based on big data alone can be unreliable,” he notes. Kenny encountered the limits of big data during his time running the Weather Company. “We did a lot to improve forecasting using big data, but when it gets serious, you deploy weather balloons and fly hurricane hunters to the center of the storm,” he says. In his playbook, the panel helps ensure that big data presents the audience’s true face.
The panels also cover a blind spot in big data. People who don’t have cable or broadband, and watch broadcast TV via an antenna, still account for one-quarter of all America’s TV viewing time. This group is heavily weighted to African Americans and Hispanics. “If you didn’t have the panels, you’d be undercounting the likes of Telemundo and Univision,” says Kenny. It’s the panels that for now are by far the best guide for advertisers seeking to reach these crucial demographics.
The rise of streaming, alongside Nielsen’s undercounting debacle, has strengthened its challengers. And those warriors at the gates have backing from the traditional networks. The group includes Comscore, 605, TVSquared, and VideoAmp.
In mid-January, NBCUniversal announced an agreement with iSpot.tv to measure audiences for this winter’s Olympic Winter Games and the Super Bowl. In an August blog post, Kelly Abcarian, NBCU’s EVP of measurement and impact, stated, “It’s time to declare measurement independence,” and that NBCU was building a “broad network of trusted partners” that could include Nielsen. In response to questions from Coins2Day answered over email, NBCU’s president for advertising and partnerships Krishan Bhatia noted, “We know that multiple audience metrics work. And in today’s media landscape, it’s the only approach that works.” Bhatia added, “For us to reach our vision, we cannot continue to bring across legacy thinking.”
Paramount Global (formerly ViacomCBS) recently reached its own deal with VideoAmp. The networks make it clear that many of these new efforts are early-stage projects. They’re testing whether these data purveyors measure a digital-migrating audience better than Nielsen’s method. Still, says Morgan, “the camel’s got its nose under the tent.”
For decades, TV measurements had a sovereign currency: the Nielsen ratings. They’ve been to buying and selling airtime what the dollar is to buying and selling cars or groceries. The networks believe a multi-currency world is probably the best solution. But they’ll need their customers, the advertisers, to accept that view. And the P&Gs and Coca-Colas that pay to run those commercials may not see value in a world where they have to sift through lots of different audience counts. “They don’t want to deal with double counting, which we take care of for them,” says Kenny.
At ad agency giant Dentsu, two top executives view the experimentation with different measurement systems as a good thing for clients. For Dave Sederbaum, head of video investment, and Brad Stockton, SVP of national video innovation, finding the appropriate alternative to what has been a Nielsen monopoly is the ideal. “It’s the best time to find the best measurement solutions from multiple providers,” says Sederbaum. “How does iSpot.tv calculate age and gender versus VideoAmp or Comscore?” Still, they note that a new currency will work only if it’s universally accepted by the advertisers, the networks, and the agencies. “We have to see where the dust settles,” says Sederbaum.
Morgan of Simulmedia believes the challengers will have an important role no matter what. They’ll be able to sell networks and streamers highly targeted secondary products—for example, delivering an audience of folks likely to file their taxes online for a TurboTax ad, then verifying that they had used the service. The secondary providers could get, say, 5% of the ad revenue for nailing those subcategories of consumers.
Still, Morgan thinks that Nielsen will prevail as the governing currency for two big reasons. “First, it’s hard to unseat an incumbent, especially when the industry’s been doing things one way for 40 years,” he says. The second reason: Morgan believes that Kenny will be able to build a better mousetrap, thanks to his deep understanding of what digital advertisers need, and his background in cloud computing and A.I.—all the products of one of the sector’s more remarkable résumés.
Kenny grew up in Lansing, Mich., son of a janitor father and payroll clerk mother, both employed by the public school. Because of his blue-collar background, his guidance counselor advised his parents that applying to college would be a waste, Kenny recalls. But winning a science prize secured him a scholarship at the General Motors Institute, now Kettering University. GM sent him to Harvard Business School, which led Kenny to senior advertising roles at the automaker and, eventually, a consulting job at Bain & Co., where he recognized how online booking would transform the travel industry.
In 1995, that insight led Kenny, then 35, to join an old-line ad agency that he rapidly transformed into Digitas, one of the first movers in digital advertising. In 2006, Kenny sold his creation to Publicis of France, one of the largest global agencies, for $1.3 billion. He stayed on to build its online franchise; in three years, Kenny more than doubled Publicis’s digital revenues to 1.5 billion euros.
In 2010, Kenny became president of Akamai, where he learned the power of cloud computing, and two years later took charge at the Weather Channel. He took that franchise from what was primarily a TV network that also offered local forecasts to a mobile app that provided targeted reports in 2.3 billion mini-geographies. “We did it by moving all the forecast computing to the cloud at huge savings, and using machine learning that kept getting better and better,” he says. Kenny separated the TV and data businesses, and sold the latter to IBM in 2016, reportedly at a huge profit to the owning consortium that included NBCUniversal and Blackstone. After a stint at IBM, he took the Nielsen gig.
Publicis chairman Maurice Lévy, an ad-industry legend, has high hopes for his protégé. “Nielsen was in a desperate situation when David came,” Levy told Coins2Day. “He can make a big difference.”
Kenny’s brainchild for the first audience-counting metric spanning traditional TV and streaming is called Nielsen ONE. Like the Gauge's platform, ONE’s methodology will combine big data and modeling with the panel review that Kenny considers such a competitive advantage. But the service won’t be available to advertisers until the end of this year, and it won’t become Nielsen’s exclusive currency until 2024. That long timeline gives advertisers plenty of time to examine the solutions proposed by Nielsen’s rivals.
Why is the rollout taking so long? The major reason is the huge gap between the way digital and linear TV sell advertising. In broadcast and cable, the measures are still based on “appointment viewing.” The ratings reflect the number of minutes the audience is watching the “pods” of four or five ads that run for two of every 15 minutes on, say, a primetime police show between 9 and 10 p.m. On Thursday evening. It’s called “selling content.” For the pod to qualify as a minute of viewing, the viewer must watch at least 30 seconds of each of the two ad minutes.
But digital platforms such as Hulu, Roku, and YouTube most often offer advertisers not individual shows, but a total audience across everything viewers watch on their platform. And they measure not by minutes that people watch the ads, but by seconds, as their viewers jump among hundreds of programming options. The streamers have data showing how many people are watching how many seconds of commercials, in an hour or a day. So an ad-supported streaming service can offer an automaker just what it wants––say, 45 seconds a day of ad exposure to each targeted viewer, every day for three months.
Nielsen ONE is now in a pilot phase, working with around a dozen partners including Disney and agency giant Magna. When it debuts in late 2022, Nielsen will still offer two different products: the traditional measured-by-the-minute offerings, and a data-rich, total audience product for digital platforms. But it will also give the traditional broadcasters the option of providing a total-audience, measured-in-seconds product. For example, a network could offer a package of 300,000 likely car buyers across its channels, unattached to specific shows or times. If they take that option, advertisers could directly compare the streaming and network products.
Here's Kenny’s big gambit. By 2025, he says, Nielsen will eliminate measurement services for scheduled, content sales, and measure ad time on cable and broadcast the same way it does on digital, by the overall seconds of ads viewed by different categories of viewers across an entire platform of programming. By that time, Kenny thinks that most of the market will have moved to streaming, including most of what’s now on cable. The networks currently complaining so loudly about Nielsen may not like the plan. But by then, they’ll be powerhouses in streaming too.
Nielsen will turn 100 about the time it introduces Nielsen ONE. Kenny is betting that Nielsen’s ability to measure media’s digital future better than anyone else will make its centennial not a memorial, but a celebration of a rebirth.
Never miss a story: Follow your favorite topics and authors to get a personalized email with the journalism that matters most to you.
