Behavioral Insider: One doesn't think of NDS in the context of advertising, or of cable TV in the context of behavioral targeting. How did all these things converge?
Gidi Gilboa: Our background has been as a technology platform provider for digital multichannel TV systems. What we traditionally were focused on was enabling cable systems both to control access to their programming and managing their relationships with customers. So what that mostly entailed was delivering program guide information, keeping track of customer accounts and usage such as what pay-per-view programming they watched. NDS is a provider of software solutions to the digital multichannel TV industry -- our software that resides at the head-end and in the set-top box (cablebox) handles all the fundamental elements of a multichannel TV service.
What become clear is that recent changes both in the market, consumer habits and technology provided a great opportunity for cable operators to leverage their relationships with consumers and the data generated by cable usage in deeper ways. With that in mind, we recently launched NDS Dynamic -- a suite of advertising solutions for set-top box that will enable addressable, measurable and interactive advertising.
BI: Can you explain the value proposition as you see it both for clients and advertisers?
Gilboa: The big bucks in television advertising have been historically going to the networks, not the MSOs. But as TV viewing habits have fragmented and changed through such things as time-shifting, the traditional advantages of network scale have eroded. Reaching a target audience has become harder and the level of accountability had become lower.
At the same time, the potential for enhanced targeting, interactivity, addressability and accountability has increased. The challenge is to enable operators to utilize the existing infrastructure, such as DVRs to enable addressable audience measurement infrastructure that already exists through set-top and DVR technology to enable targeted household level advertising.
BI: How are you adapting your platform to fill that gap?
Gilboa: We know how to push personalized ad content to set-tops through what we call live ad substitution, which involves taking pre-loaded advertisements and delivering them based on household data. A basic example would be if two different households were both watching the same program live and there was an auto ad. One household would see an ad for a Ford pick-up truck, the other would get an ad for a Ford Suburban Sedan.
BI: How would that work, and what sort of profiling is involved?
Gilboa: The ads are pre-loaded onto the hard drive. And then basically what we do is provide profiles to the box-top and use targeting algorithms to decide which advertisements to push and which ads to show (pull from the hard drive) in real time.
Initially the profiles are demographic, derived from subscriber data such as zip code. But other, more behaviorally oriented targeting is becoming entirely feasible. A simple example would be the time of day a consumer is watching. What the system does is push ads to a customer's DVR in advance and then deliver them based not only on what but when they're watching. So, for instance, if a program running live during the week contains an ad for a Best Buy sale on the following Saturday, but the customer actually watches the program on Sunday after the sale is over, the system will replace the original ad with a different Best Buy ad.
Another application is to use audience measurement to better understand how advertising is consumed. An example is coming from a commercial service operated by Sky Digital in the UK utilizing our NDS Dynamic audience measurement software. The software tracked viewing behavior of DVR users. Sky wanted to know what the real impact of DVRs was, and they found out some surprising things, [like the fact] that nearly half (44%) of time-shift ads are watched as live without being fast-forwarded. They also discovered that in aggregate DVR users actually watched a net total of 3% more ads.
BI: How granular is the segmentation, currently and potentially?
Gilboa: We can generate basic behavioral profiles. We can identify a 'Sports Fan' segment, for instance, based on identifying someone who subscribed to a sports package and also watches sports channels more than two hours a day. Though at this point using more granular sub-segments is limited by the scale of the audience base profiled, in principle we can derive very targeted niche micro-segments.
We're also able to fine-tune ads. For example if I have DVR data from a household, I may know they are from a suburban location and have a better than average income. But by analyzing patterns of viewing behavior we can identify that teen-age programming is regularly consumed at certain times of the day. Rather than deliver, say, a financial services ad based on the family household profile, more youth-oriented spots can be targeted at specific times.
BI: What kind of potential do you see with ad and content optimization in digital TV?
Gilboa: We can also learn about consumer preferences for particular types of ads by analyzing which kinds of ads a TV viewer skips or fast-forwards. This kind of data obviously will be important in targeting ads better based on individual user preference.
But its usefulness could be broader than that. You can compare engagement across groups, analyzing, for instance, whether a particular category of households are tuning away from an ad. It may also be possible to compare viewer responsiveness to different ads or even spot lengths. Over the next few years a wide range of options for testing and optimization of television will open up. These and more are all going to be possible -- while making sure consumer privacy is not affected.