Mind-reading ads

Think contextual advertising on Google is creepy? You'll hate what's coming next.

As long as there have been marketing and marketers, consumer data have been prized commodities. The Internet, which allows companies to collect information that they never could before, has raised the stakes and expectations higher still. The next frontier: combining what marketers know about customers in the offline world with what they can learn about them online, and then serving that back to those customers in the form of targeted advertising that anticipates their future wants and needs.

Call it predictive analytics 2.0. In the offline world, this area of marketing science — which involves interpreting information about customers to determine how they will act in the future — is already well developed. Telecom companies have been eager adopters, particularly when it comes to retaining subscribers. Gareth Herschel, a research analyst with Gartner in Stamford, Conn., cites the following example: “If I know Customer A very frequently called Customer B, and Customer A just cancelled their phone service to switch to a competitor, I can now factor that into my assessment of whether Customer B is still loyal,” Herschel says. The company could then offer new incentives to prevent defection.

Online marketing, despite all its bells, whistles and lightning-fast processors, is not as smart. The main reason? The technology only knows what you’re doing while you’re online, not when you’re off. Say, for example, a customer spends time browsing television sets on a retailer’s website, but doesn’t buy one. The next time that user returns to the site, they’ll likely see targeted ads promoting TV sets. But what if that customer, in the interim, already bought a television from the retailer’s physical store after looking online? If a retailer knew that before thecustomer returned to the website, it could be ready with ads not for TVs but TV accessories, such as a Blu-ray player.

The obstacles, experts say, aren’t so much technical as they are economic. The costs and complexity of capturing, analyzing and integrating online and offline info are high. But they’re coming down.

As they do, marketers will also look beyond customers’ own individual behaviour for predictive cues. A big source of valuable material in this regard will be credit-scoring agencies. “They really do know a lot about you,” says John Nardone, CEO of [x+1], which helps companies implement predictive analytics online. “They know what you paid for your house, what car your drive, when your lease is up, and so on.” The data have traditionally been used for direct-mail campaigns, but Nardone says agencies are starting to make demographic profiles of individual neighbourhoods available to online marketers.

Depending on how accurately a user’s IP address can be traced to a geographical location, marketers can use the neighbourhood to create a profile of the user: age, income, credit score or even the likelihood of children in the household. That way, marketers will have a much better idea if someone searching for toys is a parent or shopping for a friend’s kid, say, and display the appropriate ads and promotions.

If there’s any consolation for people who find such analysis unnerving, it’s that tracking an individual user — not just an IP address — to create the ultimate in targeted advertising still faces huge technical and legal barriers. “Whenever I read about these fears that we’re going to infer everything about people,” Nardone says, “I just go, ‘Oh, God. If only it were that easy.’”