Tuning in to TV data

This post is the latest in an ongoing series on The Power of Measurement. Previous topics have covered ways to make your website as successful as possible through tools such as Analytics and Website Optimizer. -Ed.

What if the ads we saw when watching TV were always just what we wanted to see? Well, we believe it is possible to make TV ads more relevant to viewers and to deliver more value to advertisers.

Television is becoming more like the web. Just as users click with their mouse to choose what's most relevant to them on the web, viewers send signals about what they want to see on television with clicks of the remote control.

Each week, Google analyzes data from millions of anonymized set-top boxes (STBs) to see which channels they were tuned to second by second. This data is provided by our partner, EchoStar. We're then able to use tuning metrics to provide our advertisers with next-day reports of how many televisions showed their ads nationwide and how the audience responded with their remotes.

We look at the various tuning metrics as signals from the audience about what they want to see and when. One of the metrics we've been exploring is the % Initial Audience Retained (%IAR). This is the percentage of the audience that was present at the beginning of the ad and then stayed tuned-in through the entire ad. If most viewers see an ad they like and decide to stay tuned-in, that ad would have a high %IAR.

Many factors affect audience behavior, including the nature of the programming, the time of day, the day of week, and, of course, the personality of each viewer. But ads themselves also have an impact. By identifying which factors affect tune-away, we can focus in on how the audience reacted to the ad itself.

Check out this video to learn what we found:



The chart below shows all TV commercials that aired on the Google TV Ads platform August through November 2008. Each dot represents an ad, and they are lined up from left to right in order of their %IAR as compared to what we'd expect given other factors (e.g., time of day, network, etc). The red dots on the left represent ads where more audience tuned away than expected. The green dots on the right represent ads where more of the audience stayed tuned than expected. The black dots in the middle are "normal," meaning there was no significant difference between the audience retention for those ads versus what you would expect based on historical data.

(Click on the image for a full-size version)

The next question we wanted to answer was how well this historical data could predict the future audience reaction. If we can use the past to predict the future, then we can get closer to putting relevant ads in front of TV viewers. So we selected one ad with relatively high audience tune-away (red dot) and one ad with relatively low tune-away (green dot) to run side-by-side on national television to test our findings. In the graph below, the diagonal line shows where audiences reacted the same to both ads. The points above that line represent airings when more of the audience stayed tuned to the ad that had previously retained audiences better. We learned audiences reacted predictably to the two ads.

(Click on the image for a full-size version)

Through our analysis of tuning data from millions of set-top boxes, we're getting closer to matching the right ads to the right television audience. It takes a lot of processing power to make sense of the enormous amount of data, but the insights to be gleaned are very powerful. Not only are we able to offer advertisers better measurement and more accountability for their TV campaigns, our goal is to also create a better viewing experience for TV audiences by showing viewers what they want to see.