InteracTiVoty: The Real Problem with Set-Top-Box Data: If You Had It, What Would You Do With It? - Todd Juenger - MediaBizBloggers

In case any of you have been hibernating for the past year or so, many of your friends in media research are becoming increasingly involved in a frenzy of interest, hope, and speculation about set-top-box data. As a guy who makes his living selling research derived from set-top-box data, that makes me happy.

Most of the current energy seems to be focused on technical measurement and editing/crediting processes, and comparing across data suppliers and research vendors. Extremely important stuff, no argument from me. I have the privilege of a front-row seat at TiVo, the only company I know of that is a set-top-box hardware manufacturer, software developer, and research provider. To my knowledge, TiVo stb's are the only ones in which data collection was prioritized and engineered into the original design. And we control all aspects of the product along the way. And it's still been a lot of work over the years to get it right. Other stb's were generally not designed to be audience meters, and bolting that capability on as an afterthought is an incredibly arduous exercise, requiring the coordination of many disparate stakeholders, all with different priorities. I could write a 3-page blog on 'things to watch out for in set-top-box data' (in fact, maybe I will). So, yes, understanding the differences in the data -- especially the veracity of the source data -- is extremely important. "Garbage In, Garbage Out" is definitely in play.

But imagine for a minute, what would happen if all of that were solved and the industry was presented tomorrow with a wealth of reliable, representative, second-by-second set-top-box data? I'm not sure anybody would know what to do with it. (By the way, same situation applies, I believe, to addressable TV advertising). Which is expected, since very little reliable set-top-box data exists today, not much thinking has been done. But it is often frustrating, since the first instinct of most anyone exposed to our TiVo data is to apply it to the same tired frameworks that have existed for 40 years. "We have one existing source counting up viewers, now we have a new one, let's compare." Nothing wrong with that, but often that's as far as it gets, and that's a travesty… but also an opportunity for the great minds of our industry to build a whole new level of understanding around marketing communications.

For example, think about reach and frequency for TV advertising. Nothing has really changed in those metrics, and how they are input into marketing plan and media mix models, for decades. But in a world where you can measure commercial audiences second-by-second – what constitutes a commercial impression? Does a viewer need to watch all thirty seconds of a :30? Fifteen seconds? Three seconds? The consumer behavior hasn't changed, people always watched parts of commercials, but we never could measure it. In fact, commercial impressions for the most part are still counted using program audiences, or more recently average "C3" audiences, not specific commercial ratings or fractional commercial ratings. And I can tell you first hand that audiences for a specific commercial can vary 3x (up and down) from the average commercial audience within a program episode. Just think of all that noise/error built into all these models. Do the differences "wash out" across billions of grp's? Maybe, to some degree. But for any specific brand, not so much.

Now take it a step further. What is the difference in the value of an impression for a viewer who saw thirty seconds versus ten? A viewer with "opportunity-to-see" in Live mode, versus a viewer who chooses to not fast-forward a spot in Timeshifted mode. Ten seconds of impression at the beginning of the spot, versus ten seconds of impression at the end of a spot? For an auto manufacturer, the value of an impression in a household that has indicated they intend to purchase a new car in the next 12 months, versus a household who hasn't? For a movie studio, the value of an impression in a household who goes to the movies, on average, twice a month versus a household who goes twice a year?

Now take it a step further. What does aggregate fast-forwarding behavior tell us about the resonance (or "engagement") of a commercial audience among the program audience? If you take the same spot, and in one program it is fast-forwarded by 40% of the Timeshifted audience, and in another program it is fast-forwarded by 80% of the Timeshifted audience, which audience is receiving the message better, and what impact does that have on effectiveness?

And now let's take it even further, beyond just television viewing. Set-top-box data finally gives us the ability to see single-source, cross-media consumption with enough scale to provide meaningful measures (TiVo and Quantcast have an offering, available today, using 35,000 opted-in households). So we can take all of the insight discussed above, from the television side, and add Internet consumption and begin to understand the impact of exposure to different marketing messages, with different frequencies, and different recencies, on different media platforms.

In the next ten years, some person, or people, is going to re-write most of what we know about marketing science, and it's going to be on the basis of set-top-box data. These people will have awards named after them, sell millions of books, and probably have a helluva lot of fun. Do we need to crawl before we can walk, and get all the technical basics of stb data worked out? Absolutely. But hopefully some ambitious souls will also start now, with the great data that's already available, to start building the foundation of the next generation of marketing research.

Todd Juenger is VP and General Manager, Audience Research and Measurement for TiVo, Inc.

Read all Todd’s MediaBizBloggers commentaries at InteracTiVoty - MediaBizBloggers.

Follow our Twitter updates @MediaBizBlogger