Editor's Note: "Genome Decoded" is a series that tells the story of Genome, our data-driven audience-buying solution, from many angles, to bring you a little closer to Big Data.
Ever see those cardboard cutouts of movie superheroes propped up in theater lobbies? That's the kind of one-dimensional view of consumers that most targeting solutions provide---thin on data and insights. Genome from Yahoo! crunches mountains of data to build rich, robust three-dimensional views of consumers and create custom target audiences that lift campaign performance.
We asked Michael Katz, Yahoo! vice president of Optimization and Analytics, just how Genome does it. Katz should know---he was co-founder of interclick, acquired by Yahoo! last year, which played a vital role in bringing our data-driven Genome audience-targeting solution to life. Here's what he had to say:
Yahoo! Ad Blog: What's all this talk about dimensions? Sounds like we're in the Twilight Zone.
Michael Katz: It's simple, really. We use a broad range of data to paint a complete picture of consumers. Most audience-targeting solutions were built to look at only one dimension, like what someone read or where they clicked. Obviously, people are multi-dimensional and audience definitions should be based on diverse data.
To reach peak performance at scale, you have to build audience models with large data sets from a wide variety of sources. We're essentially predicting which users are the best matches for a particular campaign based on what we know about them. And the more we know about them through this rich data, the more accurately we can match brands with the appropriate consumers.
YAB: How do you decide which data points are most important to helping a campaign reach its goals?
MK: It all starts with the client's goal — building awareness or increasing purchases, for example — and analyzing data on the consumers that are meeting those goals. We determine which data points are most commonly prevalent across that group of people, and our technology builds audience models of consumers who share the same attributes. By having more data to create more accurate models, we're pretty confident that our speed to success greatly outpaces any competitive solutions.
YAB: How do large data sets help you build more scalable solutions?
MK: The sheer amount and range of data we have allows us to build more complete models, which in turn allows us to build scalable campaigns, using a much more systematic and flexible approach to audience targeting.
Think of it in terms of a funnel. The bottom portion represents everybody who has already converted — that's the smallest volume, but it's the best definition of the core audience. Moving up the funnel, we find consumers who are likely to convert because they've, say, gone to the marketer's website; this represents audiences that can be retargeted. These audiences always perform the best because they've been pre-qualified and are most similar to consumers who already converted.
Next are consumers whose behaviors tell us a lot about their interest or intent, such as visiting a third-party site like a comparison-shopping engine. Finally, at the top of the funnel, we believe we've identified the most interesting opportunity, which we call "intelligent prospecting."
Prospecting, of course, is all about finding new customers. In traditional marketing, you may invest in branding and sponsorships to drive your prospecting efforts, but those are pretty broad swipes based on very little data. The opportunity to intelligently prospect consumers based on large amounts of data, even when those consumers have not yet shown any interest or intent, means marketers can confidently make intelligent investments on targeting upper-funnel consumers. This is a game-changer.
YAB: Are there certain types of data that work better than others?
MK: Our mantra is that there's no silver bullet when it comes to data. We take a holistic approach to data and look at each campaign through the eyes of the client. We partner with data management platforms such as BlueKai to acquire first-party data from mutual clients, augment that data with Yahoo!'s proprietary data, and aggregate more third-party data than other solutions to build what we believe is the most complete data set available. That allows us to create custom audiences for each campaign.
For example, take an audience like "soccer moms." Most solutions have a very rigid definition of a soccer mom, like someone who has visited a certain type of content so many times over the past month. Our definition of a soccer mom is completely fluid and dynamic, and we optimize it for every marketer, because every campaign is different. We may build a soccer-moms target audience for P&G based on 50 unique data points, but if we build soccer moms for Unilever, we might use very different data points.
YAB: Can anybody else build 3D consumer views like Genome?
MK: To do this most effectively, it takes a rare blend of the right ingredients—data, media, technology, and human capital. We've talked about our data and media capabilities, which we back with great technology. Beyond that, people and organizational alignment are key; if you don't have optimized workflows and processes, people with the right skill sets, and you don't incentivize those resources appropriately, you won't achieve full potential.
As good as our technology and capabilities are, I think we've solidified our competitive advantage with significant investments into operational innovation and organizational alignment. We've put lots of focus and effort into that over the last several years, and it's helping us deliver better solutions and build closer partnerships with our clients.