NEW YORK (AdAge.com) -- What does Google know about TV? Well, that women watch more ads than men; that about 5% to 15% of TV audiences flip channels during the ads; and that virtually no one records cable-news programming on DVRs.
None of these revelations will set the TV business afire, but they're among the first insights Google is drawing after analyzing more than a year of from 13.7 million Dish Network set-top boxes.
Google has had a tough time getting Google TV off the ground, but it was able to sign up Echostar fairly early on, in April 2007, and now has more than a year of viewing data. Since then, Google has signed up some NBC Universal cable networks, Hallmark Channel and A&E, but Echostar is its biggest trove of national viewing data across all networks.
Results of the research
Google TV showed Ad Age some of its research from the past year. The viewing data come from a subset of Dish subscribers described as "millions"; the DVR data come only from Dish Network subscribers who have DVRs.
Among Google's revelations: breaks scheduled on the hour and half-hour retain far fewer viewers than those in between; some networks retain viewers during the ads far better than others; good and bad ads follow consistent performance patterns across networks; and people consistently avoid bad ads.
The biggest audience drop-off for the ads is during live sports, when ads come during scheduled breaks, such as half-time, and during time-outs.
The data are directionally consistent with Nielsen Media Research, as well as other sources of set-top-box data such as TNS, Rentrak, TRA and the cable networks themselves. "Second-by-second data can provide important learnings into how to program pods better, how to produce better content for programs and ads, how to schedule ads to optimize commercial ratings," said Debbie Reichig, a former NBCU and Court TV research exec who is president of In-Focus Media.
A quality score for TV ads?
What will Google do with the insights? Besides share them with clients, Google TV engineering manager Dan Zigmond said, Google could conceivably use the data to come up with a quality score for TV ads, much as it does for keyword advertising, allowing TV networks to adjust pricing based on a particular ad's performance.
"In the long run, we hope we can use this data to help advertisers understand how their creative appeals to different demographics, where they're resonating and where they aren't," Mr. Zigmond said.
Google also focused its research on DVR use patterns. There, too, the research tends to align with current beliefs about TV viewing. Specifically: The same few top networks monopolize the bulk of live viewing; small, niche networks get a higher proportion of their audience on DVRs than the big networks; and long-tail networks (outside the top 100) have more people watching on DVR than on live TV.
Is this useful? Possibly, Ms. Reichig said, particularly for niche cable networks too small to be accurately measured by Nielsen's sample. "There is an important place for second-by-second data that is complementary, but it doesn't replace traditional measurement," she said.