Over the past few years, the advertising business as a whole has moved from a mood of hostile skepticism toward the Internet to an almost euphoric embrace of its possibilities. In my new job as manager of political advertising for the Microsoft Network, I'm experiencing what feels like a warp-speed repetition of the same process. For the moment, buyers are still resistant to the new medium. But I suspect that their doubts will melt as the 2000 campaign demonstrates the open-ended potential of political advertising on the Web.
The basis for skepticism about Web-based candidate advertising is that few have ever seen such a campaign, let alone one that had a demonstrable result. Peter Vallone, the Democrat who ran against New York Gov. George Pataki last time around, was one of the few candidates to make significant use of banner ads in 1998. Vallone ran negative ads about Pataki on the New York Times Web site, among other places. While one study suggests that these ads diminished Pataki's favorability rating among those who saw them, the negligible boost to a losing campaign didn't quite make the case.
A year from now, we'll have better data about the effectiveness of candidate ads. What we can say in the meantime is that Web advertising is capable of doing things no other kind of political advertising can. Here are some of the advantages I'm touting to political consultants and campaign managers this election season.
When you advertise on television or radio, you base your spending decisions on viewer and listener surveys, but you can't target precise groups of swing voters that might matter to you. With the Web, you can strike at them surgically. There are thousands of niche sites on the Web as well as Internet service providers, network portals, and free e-mail services. These sites pride themselves on collecting highly specific information about users, through registration, subscription, and the use of "cookies" (for more on cookies, see thisSlate article by Michael Kinsley). Banner ads can be directed at precise demographic groups defined by age, ZIP code, income, and various other characteristics. What's more, several of the larger portal sites are able to "merge and purge" their user data with voter lists. This means that Bill Bradley can target not just middle-aged basketball fans but also middle-aged basketball fans who are registered Democratic voters in New Hampshire. Of course, the more precise the targeting, the more expensive the ad. Reaching generic New Hampshire citizens might cost $20 per thousand impressions. For a specific category of registered voters, the price might rise as high as $70 per thousand.
The Web can serve as a tool to motivate those that are passionate about issues. A recent campaign hosted by the Juno Advocacy Network shows this potential. Juno sent an e-mail on behalf of Heritage Forests Campaign to more than 1 million subscribers. In response to this, 171,000 users sent e-mail messages to Vice President Al Gore asking him to help America's forests--without cutting down any trees. That's a 17 percent response rate. With more expensive conventional direct mail, a 4 percent to 5 percent response is considered highly effective.
Data returned from user interaction with Internet ads far outweighs what you can discern from television, radio, or direct-mail campaigns. Thanks to the miracle of cookies, clients know precisely how many of the individuals they targeted interacted with their ad in a number of ways. A banner ad that contains a streaming video version of a 30-second spot can tell you not only how many people viewed the commercial but also where in the commercial they got bored and clicked out.
The first banner ads were relatively obvious appeals for a "click-through." Now they are becoming interactive in ever more imaginative ways. Think of a banner ad as the Internet on a bumper sticker. It can do just about anything with video, sound, or animation. But first it has to grab your attention from a three-quarter inch space on the computer screen. The most popular forms at the moment include: