Don Marti

Fri 02 Oct 2009 08:47:15 AM PDT

Web ad targeting: can customers get a better deal?

Greg Sterling points out a survey that shows people prefer not to get "targeted" ads. Why not? It seems counterintuitive. Why would you want to waste your time looking at ads for things you're never going to buy? Wouldn't it be better if some nice big machine would watch you, figure out what you're really interested in buying, and only show you ads for that?

Behavioral economics aside (maybe you don't want to be tempted to buy something that you use as part of a bad habit) in real-world business negotiations, the side with better information gets the better deal.

If example.com knows that Joe rides a bike, it can send him a direct mail piece. If the company has never heard of Joe, it has to reach him some other way. The company could make Joe a better offer: a better bike, a lower price, or both, and wait for positive reviews and word of mouth that make their way to Joe. Maybe it'll get on Cool Tools. Another way to reach Joe would be to connect its offer with something that will attract riders. For example, the company could buy an ad in a bike magazine. The difference between the ad and the direct mail piece is that the ad helps fund some magazine articles that Joe wants to read, and the direct mail piece doesn't. If Joe is invisible to example.com, and the company buys an ad instead of using direct mail, he's better off by however much the articles funded by the ad are worth to him.

If example.com splashes out and buys the most expensive direct mail campaign in the world, none of that budget ends up helping Joe. If the company buys the cheapest ad in the world, that helps pay for something Joe wants to read or watch.

Today, web advertising can work like direct-response database marketing, like conventional advertising tied to content no matter who reads it, or most likely somewhere in the middle. The technology behind web advertising can make it either well-enough targeted to constitute a form of database marketing, or just attached to the content. There are ad networks that will serve the same user the same ad across multiple content sites, and site ads that are the same for every visitor. Search advertising can be targeted to the user or dependent only on the query.

When an ad network can target by user, and do it well, the best ad buy becomes the least expensive content that the user is willing to read at all. Warren Lee explains: "Through sophisticated analytics and advanced targeting/optimizations, some ad networks can deliver attractive audience demographics at scale. This aligns with the intended goal of most advertisers' media plans. As long as the content is proven to be 'brand safe,' the exact category of the content often is secondary to demographic profile for many advertisers."

Direct response advertisers have a saying: "List, offer, package." The most effective way to make a sale is to find the right people to send a message to. Second most effective is to make them a better offer. Finally, how you convey the information is least important. The less well that "list" works—the harder or more expensive it is to get information on the audience—the more incentive the advertiser should have to invest in one of the others.

Signaling value of the "package"

An ad is not just "package"—it's an important economic signal. The most useful information that an ad can convey is the fact that the advertiser is confident enough in the product's reputation to spend money advertising it. If you pick up a newspaper at the newsstand, and see an ad, you know that the advertiser spent a lot. The advertiser is signaling its belief that the product is high-quality, and its intention to stand behind the product. But the better that targeting gets, the more signaling value is lost. A banner ad could be part of a targeted buy that cost the advertiser relatively little to reach you and a few others. The more efficiently that an ad is targeted, the less information it carries to the potential customer.

Targeted advertising works against customers in two ways: first, like other database marketing, it's an alternative to making the customer a better offer, or sponsoring a better "package" of content or services. Second, it reduces the signaling value of advertising, and hides valuable information about the advertiser's intent. For example, in the IT market, vendors use advertising to signal their intent to continue supporting a product and releasing new versions and compatible add-ons. If a vendor could target only existing users of a no-longer-maintained product, it could send a false signal.

For two reasons: to get better signal and to get better content or a better offer, it should be in customers' interest to disrupt or spoof ad targeting. There's a possible example in the pharmaceutical market. Because patient privacy laws make it hard to traffic in data about people based on their medical conditions, DTC pharmaceutical ads have to be broadly aimed, and attached to expensive content such as network TV news, high-production-value shows, and glossy magazines. Advertisers can't buy a list of people with a certain medical complaint, so they have to rely on offer and package. At the other end of the spectrum is the plummeting quality of niche web sites that use sophisticated ad networks such as DoubleClick. Advertisers would rather pay for better targeting on the user than better content, and in general, it shows. (The Internet is making writers' lives harder financially, because with sufficiently good targeting, publications can profitably run ads on lower and lower quality content.)

So why do customers do so little disruption and spoofing of database marketing, and so little web ad blocking? For two reasons: many ads aren't targeted, and targeting isn't perfectly accurate, so even targeted ads still have some signaling value. And no one customer can get much of an effect acting alone. If one member of the audience decides to start spoofing, he or she doesn't get a measureable increase in content quality compared to the time spent.

If everyone in the USA agreed to recycle, unread, all unsolicited cooking and decorating catalogs, we'd all get a thicker and better Martha Stewart Living, and better household web sites, as food and housewares marketing budgets moved from direct response to ads. The web equivalent would be users starting selective web ad blocking, and accepting ads only from sites that don't offer user-based targeting, using something like the "fair ad blocking" approach.

You could look at the web ad privacy bill from Rep. Rick Boucher as a way to break the stalemate, and use the law to help customers by shifting budgets away from database marketing to better offers and content. The law is the wrong approach, though—all that CAN-SPAM did was encourage spam by overriding state laws that let recipients sue spammers—but it is worth looking at technical ways to help meet users' demands for less effective tracking. Would ad blocking tools be more successful if there were a blocklist that only blocked ad networks with sophisticated targeting by user, and let the rest through?