Have you ever been followed around the Web for months by ads for a product that you considered (but decided against) buying from an online retailer? Have you seen ads on Facebook offering some product that a marketer thinks a person like you should need and want?
Many people in the Web advertising community seem to believe that consumers will appreciate this kind of personalized targeting, because we’ll only have to look at ads for stuff we’re interested in. For many of us, though, that isn’t the case at all.
Behind the Web that’s visible to us lies a massive and complex advertising machine that wraps virtually every kind of content we see--from mobile games to Web searches to viral videos--in ads. Some of the older machinery may be familiar to you, but some of the new stuff still lurks in the shadows, evidenced only by its product. In this article we’ll look in rough outline at some of these advertising systems and how they work, starting with older technology and ending with newly minted stuff that’s just coming into use today.
Retargeting (aka Remarketing)
Retargeting/remarketing involves having an advertiser drop a cookie into the consumer's browser that enables the advertiser to follow that person around and show them targeted ads for the product/retailer after they leave the original site. This practice is far from new--it’s been going on for about a decade. Many big-name retailers have adopted the practice. Overstock.com may be the most blatant: Click a product at that site, and you might very well see ads for it at other sites for weeks after your Overstock visit. Smaller Web vendors such Tom's Shoes, Blue Nile, Timbuk2 are using the technique, too. Whereas Overstock may follow you around with ads for products that you browsed or left in your shopping cart, smaller retailers just follow you around with a general ad for their brand.
Advertisers (whether businesses or advertisers representing businesses) buy leads from websites based on how groups of consumers behaved at the site. The advertiser can “cookie” the consumer’s browser at different places on the site, determining how qualified a sales prospect the consumer might be. For example, Shopper A who merely lands on the homepage of a site (and goes no further) gets one kind of cookie; but Shopper B who lands on the homepage, browses products, puts something in a “shopping cart” and then proceeds to “checkout,” receives a different cookie. Shopper B is more valuable to the marketer than Shopper A, because B exhibited browsing behavior that suggests real product interest and a greater likelihood of actually buying something.
A retailer may also drop cookies based on specific products that a consumer browsed at the site. If the consumer doesn’t buy, the retailer may pay Google or Bing to target ads for that product at that user. “If you have a category of products that you make more margin on and someone lands on this page, then you may be willing to pay more to get them back as a customer and have them convert since you stand to make more money,” explains Meaghan Danielson of the Web advertising firm Adlucent.
Danielson also points out that advertisers don't concern themselves with specific consumers; instead, they “buy audiences” of consumers--thousands of advertising targets that have been grouped together based on the way they behaved online. “There is no tracking back to individual data or anything like that; generally, all of the tracking happens on the back end and is never looked into deeper than creating an audience, so privacy isn't a big issue here.”
That's not to say that it isn't annoying. One colleague told me that Groupon followed her around with retargeted ads for nearly a month, and she got sick of seeing them. Mozilla has implemented the Do Not Track standard in Firefox 4, and Adblock Plus remains the most popular extension for the browser. Most advertisers, however, limit the number of impressions that they show to certain audiences so that the ad will show only a certain number of times a day or month, and advertisers usually allow the audiences to expire after about 30 days.
“Social” marketing is the new wave. Social marketing is a much more fine-grained approach to targeting potential customers than relying on the traditional demographics approach (for example, "42-year-old men in Montana often buy Ford trucks").
This new set of advertising techniques uses “social graph” data to determine what products Web consumers might buy. The social graph refers to the highly personalized data that sites like Facebook collect. A Facebook profile may contain your age, sex, and location information. It may also know that you are an avid hiker who attended a Coldplay concert last Tuesday night. It may even know for example, that you recently visited the websites of Fox.com and RNC.com. It may know that you belong to a gay rights Facebook group.
A marketer such as Coca-Cola or Saturn or Nike could analyze this data set, group it with a bunch of other people who have been identified as having similar interests, and then release targeted ads at them. Or a marketer could determine that based on your similarity to people who have already bought its products, you are a good “lead’ and should be targeted for advertising.
Facebook’s social grid is said to be “closed.” In other words the company does not sell chunks of its huge user data set to third parties. But you don’t have to buy user data from Facebook to obtain it. Some marketers plant a widget (a small app) at their Facebook page, and then track all of the people who click on the widget. And make no mistake that the companies that post quizzes, lists, and games on Facebook are indeed aggregating the personal data of everyone who agrees to their terms of service. Web marketers can acquire user data in various forms from them, too.
Facebook also collects and makes "public" the list of people who are your Facebook friends. Marketers naturally want to reach people who are likelier than most to buy their product, so it makes sense to target people whose friends have already purchased the product. The tastes and buying habits of your circle of friends, in other words, are much better predictors of what you're likely to buy, than are your age, sex, and location data.
Bringing It All Together
The most up-to-date social marketers combine old-school digital analytics with the new world of social marketing.
Some examples: Media6Degrees has made a business of finding (and scoring) similarities in the social data of different groups of people, yielding a "social graph" of preferences. Phillips says that one of his company's clients may provide Media6Degrees with a list of its customers, and then ask Media6Degrees to find people with similar social graphs from within its database of roughly 20 million people, the assumption being that similar people with similar tastes are likely to buy similar things--that they share a "common brand affinity."
Other marketing firms, like 33Across, license the user data collected by blog sites, social media sites, and app developers. If the company identifies a segment of users who seem likely to buy a particular product, it may drop a cookie into the browsers of those users on behalf of one of its clients. 33Across can also acquire a list of its client’s current customers, and then direct targeted ads at those customers’ social network friends, on the theory that like minds buy alike (or at least similarly).