Background
This ratio percentage was not created in a vacuum. We consider it a next step in a series of careful analyses of various cuts of set top box data from different data aggregators. We present it as one more example of the efficacy and usability of set-top-box data for measurement.
Past analyses have shown that when set top-box-data results are compared to Nielsen:
advertisement
advertisement
With the knowledge gained from these past analyses, we felt that we could create a ratio percentage using STB data that can be used to help translate the performance of networks that are not measured by Nielsen into an agency's measurement parlance. What we did was to compare the delivery of a chosen network to a select affinity group of measured-by-Nielsen networks (such as Kids, Sports, Business and General Entertainment) in STB data and then apply this ratio to the actual Nielsen delivery for those same measured networks.
We examined a range of networks using Rentrak's AT&T U-Verse data, including Bloomberg, INSP, Sprout and Wealth. All these networks subscribed to Rentrak measurement. This does not include set top boxes in their Dish and other platforms. For the record, Rentrak reaches over 15 million TV sets. For the purposes of this analysis we used a subset which was the complete U-Verse universe of 2.5 million homes.
The following people were involved in discussions about the project at some point in its development: Joe Abruzzo (MPG), Brad Adgate (Horizon), Matthew Bayer (Carat), Shari Anne Brill (Independent), John Cogan (OMD), Ed DiNicola (Independent), Andy Donchin (Carat), Joan Fitzgerald (comScore), Frank Foster (AT&T U-Verse), Jason Kanefsky (MPG), Bill Livik (Rentrak), John Morse (Byron Media), Jim Multari (PBS Sprout), Mitch Oscar (MPG), Stu Rodnick (Three Screen Nation), Art Salisch (Scarborough), Jonathan Steuer (AT&T U-Verse), Rick Wardell (INSP), Charlene Weisler (Independent) and Leslie Wood (Independent).
Sprout Example
In creating a ratio percentage for Sprout, we chose three Kids prototype networks that were measured by Nielsen and that best matched the Sprout profile. Using Rentrak AT&T U-Verse of 2.5 Million households for prime time 8 p.m. -11 p.m., we examined the following weeks: Jan.18-24, 2010, Oct. 19-25, 2009, July 20-26, 2009 and April 20-26, 2009. We examined the same week per quarter to help ascertain seasonality.
Starting with Jan. 18-24, 2010, we grossed and then averaged the audience delivery for the three prototype networks. Then we calculated Sprout's percentage delivery of that average which was 9% for that January week.
We then examined Nielsen performance for the three prototype networks for the same week in January, grossed and averaged these deliveries and then applied the 4% ratio to the average. The result was 38,357 or Sprout.
After calculating January, we wanted to be sure that the ratio was stable and not subject to seasonality. In applying the methodology to the other three quarters we found that the ratio percentage was comparable.
Conclusion
While the creation of performance ratios through Set Top Box Data is still in its developmental stages, the ratios created with prototypes using Rentrak's AT&T U-Verse data has met with research and buyer sign-off as a plausible approach at agencies such as MPG and Carat. We hope to be able to examine more datasets over the next few months. This initial step indicates to us that the ratio method has the potential to create meaningful, comparable delivery impressions for certain networks in the television universe for use at the agencies.
Charlene, does this projection technique take into account "tiers"? Being from DownUnder I am unfamiliar with the detail of your tier structures, so please bear with me. Is Sprout on U-Verse on a similar tier as Sprout on other subscription-based services? For example, Sprout may be a "basic" channel on the U-Verse service and not others, which would/could inflate the ratio. Conversely it may be on an "extended package" on U-Verse and be a basic channel on others, which would/could deflate the ratio.
Apart from that the approach seems fair.
Thanks for the detailed example Charlene. It takes me back to the early days of cable (circa 1985) when MTV would do coincidentals and then we (at Y&R) derived adjustments from vpvh's that were applied to Nielsen ratings.
Hi John,
That is a good question. Even though the ratios are fairly proportional, we did not take into account differences in penetration or tiering in this first look. We will be continuing our analyses in the coming months.
We did test the ratios with networks of all sizes that are measured in both services and found that the ratio was an excellent projection technique even with different distributions between the services.
Hi Rob,
Thank you. And I also heard that magazines use similar techniques to project performance.