Balancing the need for innovation and continuity in audience measurement

Dramatic changes in TV viewing behaviors over the past decade, driven primarily by widespread connected TV (CTV)¹ adoption and increasing high-speed internet availability, have introduced an array of new data sets that are finding their way into measurement solutions. Generically referred to as big data, these data sets provide the opportunity to advance audience measurement to help both buyers and sellers better understand how TV programming—and ads—are performing. 

The scale of big data

At a high level, there are two primary sources of big data that are being used in linear television audience measurement:

• Return-path data (RPD) from cable and satellite set-top boxes (STBs)
• Automatic content recognition (ACR) data from smart TVs

With 70.6% of U.S. TV homes owning a smart TV, up from 62.3% two years ago, one of the biggest benefits of big data is scale. Scale is important in today’s fragmented viewing landscape. Our big data set currently includes 45 million households in the U.S. and 75 million devices from Comcast, Dish, DirecTV, Roku and Vizio, which rivals that of any other measurement provider.

However, these data sets are not uniform or homogenous, and they were not designed for use in audience measurement. But that doesn’t mean that they can’t be used for this purpose. In fact, they can be very useful—but not by themselves. Making sense of big data requires a truth set that corrects for gaps, fluctuations and other intricacies in the big data set.

Ensuring measurement stability & representation

Pairing big data with a representative panel is critical to account for viewing across all devices and audiences in a stable way. Nielsen has a representative panel of 101,000 individuals from approximately 42,000 households that enables us to harness the power of big data while correcting for its shortcomings. 

And the industry agrees. The World Federation of Advertisers’ (WFA) ‘North Star’ principles for cross-media measurement call for a combination of quality panel and big data. In August 2022, both the Video Advertising Bureau (VAB) and the Association of Advertisers (ANA) announced plans to build their own panels, as did Google for use in its online conversion measurement. Put differently, the industry is clamoring for approaches that both ensure representation and harness the power of big data. We believe Nielsen is uniquely positioned to provide these solutions.

We’ve invested a decade of research in integrating big data sets into our methodologies in innovative ways that also ensure continuity with our currency measurement. We incorporated big data into our U.S. local TV measurement in 2019 and into our national TV measurement in 2022. When used in concert with representative, persons-level panels, these big data sets can significantly advance the science of audience measurement.

1. Measurement is about people. Big data provides no information about the people who are doing the viewing. By pairing big data with panels, we’re able to understand who is viewing, as well as household makeup.
2. Measurement must be representative. Big data provides an incomplete picture of TV viewing. For example, RPD/STB and ACR data lack streaming coverage and over the air (OTA) viewing. As of November 2023, streaming accounted for 36.1% of television viewing in the U.S. Additionally, 18.1% of U.S. TV households have at least one TV set that accesses content using a digital antenna instead of STBs or an internet connection². Big data viewing sources also may not include all devices in the home. This is particularly an issue with ACR data where the number of devices returning data is about 1.1 per home. The average U.S. TV home has about 2.5 sets. Panels are critical to cover these gaps, and any measurement solution that solely relies on big data would miss these audiences.
3. Measurement is more than just data sources. ACR data, for example, simply identifies images on a screen. If the same piece of content is airing on multiple channels at the same time, the ACR data has no way of accurately attributing viewing to one channel versus the other. Similarly, RPD and STB data are often incapable of verifying if a TV is even on. Different RPD/STB providers and original equipment manufacturers (OEMs) also have different ways of collecting and processing data. This makes ingestion, harmonization, householding, and calibration both critical and extremely complex. 

Measurement shortcomings in RPD/STB and ACR data highlight why it is critical that these data are calibrated with people-based panels that accurately represent the diversity of a population. 

Navigating the evolution of measurement

By marrying the scale of big data with person-level information from innovative new panel meters, we’re able to provide enhanced granularity down to the individual commercial. This ensures advertisers can understand both who and how many people saw their exact ad. We’re also able to provide more advanced targeting to unlock the ability to capitalize on linear addressable capabilities, as well as advanced audience segments. 

There is no shortchanging the benefits Big Data+Panel will bring, but the shifts taking place in the media industry are far too significant to make wholesale methodology changes to data being used as currency on a moment’s notice. Stable transition in times of change will always be better than a very quick pivot. Times of transition also amplify the need for transparency, especially when measurement data underpins a global industry that ad research firm WARC estimates will top $1 trillion next year. 

Recognizing that the industry needs time to adapt to innovations in measurement, we made our national Big Data+Panel data stream available for transactions in parallel with our TV panel currency for the ‘23-’24 television season. For the ‘24-’25 season, we have enhanced our Big Data+Panel stream to include Comcast data (in addition to DISH, DirecTV, Roku and Vizio), and we’re committed to helping the industry navigate through this transition. Well aware that different organizations evolve and adapt at different rates, we will also provide our audited and accredited TV Panel data (average minute), as well as Big Data+Panel (at the average minute and individual commercial minute) during the coming season.  

The evolution of TV measurement will only happen with industry alignment. And while many agencies and advertisers are interested in moving toward measurement of exact commercials, the overall sentiment in the industry is that it’s not yet ready to retire C3/C7 ratings. We need to balance the equally valid needs for innovation and continuity. With three data sets available, buyers and sellers will be able to evolve systems to truly operationalize the planning, selling and buying of national linear TV at an exact commercial level. 

Sources:

¹ CTV refers to any television that is connected to the internet. The most common use of the internet connection is to stream video content
² Nielsen National TV Panel, November 2023