When the online display industry got started over a decade ago, advertisers were focused on buying "eyeballs." You tried to reach as many as possible for your budget, but targeting was rudimentary. To get to women, for example, you would buy an ad placement on an online women's magazine.
Today, with real-time bidding and exchanges -- and tools for analyzing petabytes of behavioral, contextual and sales data -- advertisers buy not "eyeballs" but "audiences." They go after targeted shoppers who are in the market for certain goods or are likely to be interested in a certain brand. Display-ad networks like Yahoo, AOL, Google Display Network, Audience Science, ValueClic or Specific Media provide a list of audience segments, and media buyers choose the ones they want to target . It might be "sedan shoppers" or "digital-camera buyers."
But even audience buying is not finely tuned enough in a world where customers are finicky, highly adept at searching for products, brand-disloyal and impervious to all but the most targeted advertising messages. Consumers thinking about their next luxury sedan while browsing international vacation packages are different, for example, from folks deciding between buying a Veloster or a CR-Z this weekend. Yet most ad networks would treat all of these people as "sedan shoppers."
So what's next? It's already here. Call it "artificial-intelligence targeting," for lack of a niftier term. Media buyers might describe an ad placement like this: "men aged 35-44 with two kids who live in the Buffalo, New York, area, read car-enthusiast web sites, are in-market to buy a car, and love the Audi brand –- but could be tempted to buy a Lexus in the next 2 months." But with this new technology we need not be limited to thinking in terms of a handful of attributes (say seven, as above). We can meaningfully use the tens of thousands of attributes that exist about consumers.
How is this possible? The surge in availability of behavioral, demographic, intent and other kinds of consumer data means media buyers have access to millions of real-time data points. Of course, to make sense of this ad data deluge, you need super-advanced software to crunch it, analyze it and spit out actionable metrics. Artificial-intelligence programs use advanced math to analyze the data, identify connections and adjust to patterns that emerge along the way. Each successive ad buy can be more and more targeted and effective.
As the targeting grows increasingly micro, marketers actually get more scale than they could before. Micro-targeting means micro scale if you use only 5 or 10 consumer attributes. But if you leverage tens of thousands of attributes, you will find lots more people who could be a good fit for your product or message. When we add in all the data, we increase performance (better accuracy) and scale (we find more people because we have more ways to find them).
To be effective using the new technology, a campaign will need to analyze as many of a prospect's attributes as possible. It's the difference between focusing, say, on "women age 25-44" as opposed to "women age 25-44 who buy more than 12 pairs of shoes per year, live in a sunny climate, always buy full-price, favor gladiator sandals and love Jimmy Choo." Success also requires updating campaigns in real-time as targets change.
Consumers don't think of themselves as audiences, but as individuals with personal tastes and needs. Artificial intelligence allows media buyers to move beyond buying audiences and instead buy every ad with many individuals in mind.