Inside Nielsen’s push to incorporate streamers’ first-party data into measurement

Nielsen is trying to develop a new system for incorporating first-party data from streaming networks and platforms into audience measurement, which could prove particularly important for measuring the biggest cash cow in TV—live sports. But first, it will have to get widespread agreement on how to do it, and that’s likely to prove elusive in a market where the biggest streaming players on the sell side don’t have a unified voice on how to manage first-party data. The U.S. Joint Industry Committee may have a roadmap for getting streaming data to measurement companies from many networks, but neither Nielsen nor the biggest streaming platforms are part of that. Check out the latest Ad Age events and award programs How it all ends could affect billions of dollars in ad spending and network revenue annually as more viewing, and more live sports in particular, shift to streaming. Measurement that uses panels or big data from set-top boxes and smart TVs has limitations that only reliably catching and processing first-party data from media companies can resolve. Results for the biggest live streaming event to date—Peacock’s Jan. 13 broadcast of the NFL playoff game between the Kansas City Chiefs and Miami Dolphins—point to how much of a difference there can be between measurement approaches today, given the continued divergence between Nielsen and iSpot.tv numbers. Nielsen said it didn’t use first-party data to measure that game; iSpot possibly did. Either way, an industry-accredited, standardized system for using first-party data could help alleviate doubts. The battle over sports measurement Since live sports remain the most watched programs and get the most ad revenue, they get the most attention. Nielsen executives held a virtual meeting with executives of TV networks, ad agencies and sports leagues last month to talk about measurement issues, including streaming, similar to an in-person event held the year before. While the nation’s leading measurement company offered no details on exactly how it will get and process first-party data from networks, Nielsen indicated it would be a comprehensive solution that treats all networks equitably, according to people familiar with the matter. But so far, efforts to get first-party streaming data into Nielsen measurement have sparked debate, specifically when Nielsen proposed a system to model its panel currency data using summary first-party data from Amazon Prime for “Thursday Night Football” last year. Other networks balked, including via their trade group, the Video Advertising Bureau, saying they didn’t get the same opportunity to have their first-party data incorporated by Nielsen. Ultimately, the industry’s measurement standards body, the Media Rating Council, declined to revise Nielsen’s accreditation to allow first-party data from Amazon into its accredited process. But Nielsen has made a blended panel and first-party Amazon data solution available for those who want to use it. Read more: Nielsen won’t force Amazon’s data into ‘TNF’ ad deals Nielsen so far has declined to apply for certification from the U.S. Joint Industry Committee, which includes Comcast’s NBCUniversal, Paramount, Fox and Warner Bros. Discovery, along with other networks and agencies, and is another potential avenue to getting first-party streaming data. The JIC has set out certification as a prerequisite for measurement companies to access the Streaming Data Server it’s planning to launch later this year, with Paramount being the most vocal on that exclusion. VideoAmp and Comscore received full certification last week, with iSpot.tv continuing on conditional certification as it makes changes to its processes. Even Nielsen CEO Karthik Rao, who’s been a JIC critic in the past, had some kind words for the JIC process when speaking at a Coalition for Innovative Media Measurement conference on April 3. “You absolutely do need standards,” Rao said, “evidenced by the work we did last year in football and how the industry responded to that. There were very specific things that we did and are responding to, which means there is an industry body taking accountability for how that should work. “The work that the JIC is doing in this area is a really good step in that direction, clearly, because it takes a little bit of the pressure and stress away, as long as it’s audited and follows the standards established,” Rao added. “It can’t be that every provider is stuck within the constraints of each individual streaming service, because it’s all going to be different, and harmonization and standards are going to be critically important.” Also read: How cable TV will compete for upfront ad dollars Questions about Peacock data One example of why it is so important to get first-party data applied to measurement, especially as it pertains to streaming sports, is the divergence in numbers for that first all-streaming NFL playoff game on Jan. 13 between the Chiefs and Dolphins on NBCU’s Peacock. ISpot.tv data for that game came in more than 20% lower than Nielsen’s, according to people familiar with the matter. That’s a marked contrast to the data iSpot has released over the past two years for big TV events, which consistently came in higher than Nielsen’s. Those iSpot numbers came to the attention of the MRC, which inquired about whether Nielsen did anything different than its usual panel-based accredited process for the Peacock game. Ultimately, the MRC was told Nielsen did not do anything differently, and Nielsen has said there isn’t anything wrong with its Peacock number. “The only entity raising this is a self-interested, unaccredited measurement provider that rated the game,” a Nielsen spokeswoman said. “It’s a manufactured non-issue.” Media measurement blog Tracking TV, social and digital updates Click here Nielsen’s measured audience for that game came in at an average of 22.8 million people. It was the smallest NFL playoff audience of the day, but a strong showing given that it was the first ever playoff exclusively on streaming (outside local broadcast markets). The results were lauded in a press release by Comcast CEO Brian Roberts, whose company paid $110 million for the rights, and by NFL Commissioner Roger Goodell, whose league seeks to expand streaming deals. The iSpot audience for that Peacock playoff game came in at around 18 million average viewers, more than 20% lower than Nielsen. That has surprised people who heard about the difference, given a pattern where iSpot numbers have come in higher than Nielsen’s by 2% to 13% for NFL playoffs last year and the Super Bowl the past two years. ISpot’s audience was also 17% higher than Nielsen for the April 1 Elite Eight matchup between Iowa and LSU in the women’s NCAA March Madness basketball tournament, and 2% higher for the Academy Awards on March 10. To be sure, whenever Amazon has released data that incorporates its first-party data with Nielsen’s for “Thursday Night Football,” those numbers have come in higher than Nielsen’s panel-only measurement. And Nielsen’s own use of big data modeling has raised the size of audiences across all age ranges, compared to using panel measurement alone, as Pete Doe, the company’s chief research officer, noted in a press briefing on April 2. So it’s unusual for a panel-based measure to show a larger audience than one based primarily on big data sources or first-party streaming data, as is the case with iSpot. ISpot, which didn’t release NFL playoff numbers publicly this year, declined to comment on data for the Peacock game. “It’s kind of like trying to solve a one-sided equation,” said George Ivie, CEO of the MRC, about looking into the divergence. That’s because Nielsen’s panel-based data is accredited and audited by the MRC, and iSpot has not yet sought accreditation for audience measurement. “I know what happened with the Peacock game as it relates to Nielsen measurement,” Ivie said. “It was essentially panel-only measurement plus DTVR [Digital in TV Ratings], which is an add-on digital measurement product that supplements the panel. We ended up concluding that it was derived through our audited process.” Ivie said he asked Nielsen to confirm that it didn’t use the same process with the Peacock game that it proposed to use with Amazon football games last year. “They confirmed back, no, that wasn’t the Amazon process,” Ivie said, and that Nielsen said it had followed the MRC’s guidance last year not to implement that process for Peacock. One reason for the difference in numbers could come down to a difference in how Nielsen and iSpot measure streaming. Nielsen relies on its household panel; iSpot relies primarily on big data sources plus the 5,000-household TVision panel, including smart TVs and Conviva, which gets and processes first-party data from streaming platforms. ISpot declined to comment on whether it used Conviva’s first-party data to help measure the Jan. 13 game, though Ivie said he believes that was the case. Inexact science To be sure, even MRC-accredited measurement firms can diverge on a tentpole event with the biggest audience—the Super Bowl. Comscore, which received MRC accreditation last month to measure audiences at a household level, reported in February that the Super Bowl had a 61.2 average household rating, which translates to 76.5 million households. That’s 36% higher than the 56.4 million households (43.5 rating) Nielsen measured. For streaming, iSpot and VideoAmp both use tech provider Conviva as a conduit for collecting, analyzing and cleaning up first-party data from networks in calculating audiences. Nielsen declined to comment on whether it has used Conviva data in any cases. Conviva is open to selling to all measurement companies, including Nielsen, Keith Zubchevich, Conviva’s CEO, said. Zubchevich heartily endorses Rao’s point that measurement companies shouldn’t just take any data that networks opt to send. “There’s a misunderstanding in the marketplace that you can just collect any data from anywhere, any way you want and somehow throw it into a database and it will all make sense,” Zubchevich said. But it has to be cleansed, reconciled and normalized across a range of media companies, devices and operating systems, then reported to measurement firms in a standard way, he said. Even “big data” isn’t configured to do that. Cable companies and smart TV manufacturers are restricted either by technology or licensing deals from reporting viewership numbers from streaming apps. Automated content recognition data from smart TV manufacturers “generally don’t have a lot of streaming data in there, so you have a big gap for streaming consumption,” said Deirdre Thomas, chief product officer of Nielsen, in a media briefing last week. “In Nielsen, we have that measurement in our panel.” But Nielsen’s panel of 42,000 homes, including 20,000 with streaming meters, has long been said by industry leaders to provide limited granularity in an increasingly fragmented streaming market. “The smaller the sample set, the error margins just go way up,” said iSpot CEO Sean Muller at the CIMM conference. He said he believes audience measurement among rivals actually will converge more in the future as they rely more on first-party data. “We’ve seen some initial evidence to suggest that traditional measurement has not measured streaming correctly using the same approaches it uses for linear,” Muller said. “But with streaming, with availability of first-party logs, it’s really going to become less about counting, more about verification, deduplication, outcomes, reach and frequency across platforms. It will converge when we get to a streaming-led world.” Easier said than done Of course, that hasn’t happened yet. And Nielsen’s priority at this point is getting accreditation for adding big data from set-top boxes and smart TVs to its panel-led accreditation, which it hopes will happen by May, a spokeswoman said. That audit and process do not include incorporating first-party data from networks, she said. Nielsen’s spokeswoman declined to comment on technical details of how it plans to standardize incorporating first-party data from streaming players, but said: “We are working with all clients to understand viable paths for big data integration with them.” As for industry standards around doing that, they already exist, the MRC’s Ivie said, and Amazon and Nielsen did look to work within them in proposing a modification to Nielsen’s accredited panel methodology last year, he said. Though MRC membership didn’t approve the Amazon-Nielsen methodology, Nielsen’s spokeswoman said the company will bring first-party data integration to audit separately from its current big data vote. Regardless, getting the whole industry engaged with a standard way of sharing and processing first-party streaming data is far from a sure thing. The JIC has developed standards for data sharing among four of the biggest TV-centric networks and several others. But the biggest streaming players, including in sports, such as Walt Disney Co., Amazon and Google’s YouTube, aren’t part of the JIC at this point. And Disney has no plans to join, Lisa Valentino, executive VP, advertising, Disney, said during the CIMM Summit. “The majority players in the streaming marketplace are not sitting at that table,” Valentino said. “And that’s a problem. … We have to solve for partners where the majority of their supply is addressable, data-enabled and tech-forward.” Media measurement blog Tracking TV, social and digital updates Click here

Inside Nielsen’s push to incorporate streamers’ first-party data into measurement

Featured Stories