Death of a Salesman?

Google is testing a system that seems designed to marry two powerful ad-serving technologies into an even more potent way to direct relevant commercial messages to a given web site. According to a report in MediaPost, Gokul Rajaram, a group project manager for Google’s AdSense program, said the system under development will allow publishers “to submit demographic and psychographic data about their audiences” so as to better target ads and improve response rates.

MediaPost says the new Google pilot program started about a month ago. “Many of the signals that Google will be using are demographic,” MediaPost reports, adding that “Rajaram said that publishers (will) also (be) supplying other indicators, such as information they know about their visitors’ other interests.”

It sounds to me as if Google is attempting to blend contextual and behavioral ad matching. In the past I’ve tried to highlight the differences between these two approaches, but a quick example here may suffice. An article about the Lord of the Rings might draw a contextual ad for Rings books, movies or action figures. A behavioral ad match might note that Tolkien fans tend to visit online role-playing games, and place an ad for a new game alongside the content.

Jupiter Research analyst Gary Stein told MediaPost that the latest Google move is a bid to stay ahead of competitors like Yahoo, which recently launched a publisher network that also seems to allow some blending of contextual and behavioral ad placement. Here is an excerpt from the MediaPost story on Yahoo’s move:

“A music site publisher might know that the site’s audience is interested in travel. That publisher will be able to go into a user interface and tell Yahoo! to serve travel ads to his site, or to specific pages within his site. This feature raises the question of whether the ads are still “contextual.”

I say: “Context, schmontext.” Publishers want ads that deliver results for their advertisers and revenues to themselves. Call the technology whatever you like. Having said that, I know there is some reticence to put the “behavioral” tag on ad-serving software because that implies publishers and advertisers are tracking people’s behaviors. Well, aren’t they? What is the Amazon suggestion engine but a facility that tracks past behavior and suggests future action?

What I find more interesting, in light of these continuing improvements in automated ad-matching, is the extent to which algorithms can replace the old-fashioned salesperson. Certainly for a small publisher, the ability to gain any revenue without a personnel expense is attractive. For larger publishers, these automated responses can deliver baseline revenues, which can be used to hire a salesforce. As I noted in a July posting, Weblogs, Inc. principal Jason Calacanis told Online Journalism Review he “derives the majority of his income from display ads sold directly to advertisers.”

Because an algorithm can be liked. But I don’t think it can be well liked.

Tom Abate
MiniMediaGuy
‘Cause if you ain’t Mass Media, you’re Mini Media