1. Data & Insights

Big data virtually eliminates zero ratings in national TV measurement

Pete Doe

In the U.S. and around the globe, fragmentation of viewing has challenged the ability of traditional audience measurement panels to deliver reliable ratings information. That information remains essential for ad buyers and sellers to plan and execute deals, and for content producers to understand program performance. 

Although higher-rated networks and programs are less affected by this measurement issue, a lot of viewing today is niche, leading to lower ratings and less-reliable measurement. In some cases, this can lead to ratings being reported as zeroes, where despite there being viewers to a program in the population, none of them were on the measurement panel. To address this, Nielsen is releasing its “panel plus big data” measurement data. This entails combining panel data with big data—return-path data (RPD) from MVPDs and automatic content recognition (ACR) data from smart TVs, significantly improving the audience measurement by making it more precise and effectively addressing the problem of “zero ratings.”

We know that the data from cable boxes and smart TVs isn’t feasible for measurement by itself. It has gaps and doesn’t measure everything on the screen or tell you who’s watching. But, when paired with accredited person-level panel data, the combined data sets significantly advance the science of audience measurement. Among the benefits, the combination of these data sets allows measurement to collect far more viewing than what’s possible with panel data alone.

Understanding zero ratings

In the first quarter of this year, for example, The Nielsen panel reported zero ratings for 8,454 telecasts across broadcast, cable and syndicated TV, out of a total of 362,168 total telecasts (2%), for live telecasts and the following three days of time-shifted viewing. A rating of zero means that no homes in the panel tuned in, which in most cases, reflects the limitation of the measurement rather than what actually happened in the population. In other words, there was some low level of viewing in the population, the viewers were not found in the panel.

The inclusion of big data in Nielsen’s national panel data significantly addresses this, allowing buyers and sellers to have clearer visibility into audience viewing behavior. Nielsen is combining TV data from approximately 30 million homes with its 40,000 home/100,000 person panel, greatly increasing the precision of measurement. The larger sample size reduces the statistical sampling errors associated with panel measurement, including the presence of zero ratings.

Big data facilitates greater viewing precision

Before making big data available for measurement this September, Nielsen spent over a year analyzing the combined data sets to understand the impact on sampling errors. The analysis found that the inclusion of big data significantly reduced zero ratings across all viewing groups. When the 8,454 TV program telecasts in the first quarter that generated zero household ratings were measured with panel data plus big data, the presence of zero ratings declined by 99.9%. Across all major audience demographics, the declines ranged from 96.8%-99.9%. The reductions were most notable among younger audiences, who tend to be lighter viewers of traditional TV programming and are therefore more likely to cause zero ratings in panel measurement.

The increased sample size also allows more precise viewing to be collected among specific audiences. In the first quarter, for example, there were 3,471 Hispanic program telecasts that generated zero ratings (live+3 days) on Hispanic networks when measured with panel data. When measured with panel data and big data, the zero ratings were removed entirely.

Clearly, the addition of millions of homes into measurement is positive. However, while big data provides a significantly larger sample size, on its own, it lacks the specific detail about who is watching, and typically under-represents diverse populations and certain age groups. Big data is also unable to measure viewing from over-the-air antennas, broadband-only homes and viewing that takes place away from the home. That’s where audited and accredited panel data comes into play. By combining the scale of big data and granular insights from our people-based panel, Nielsen is helping the industry realize the full potential of RPD and ACR data.

The post Big data virtually eliminates zero ratings in national TV measurement appeared first on Nielsen.

View Original Article
https://www.nielsen.com/us/en/insights/
Do you like Nielsen's articles? Follow on social!