Why I Love Display Ads
(by George John)
When I was a kid, I would spend about 20 minutes reading a Spider-Man comic book, then two to three hours looking at the ads in the back and imagining how cool it would be to have X-ray glasses or my own flying-saucer hovercraft. Today, I subscribe to the print edition of the “San Jose Mercury News” for the ads, not the articles. Nothing beats scanning eight square feet of Fry’s Electronics ad all at once or scanning the paper for local ads for theater and family events.
This Is a Good Thing…Except for the Ad Networks
I get almost all of my news for free through the Web, as can anyone with Internet access thanks to display ads. As in other channels, online display ads are a democratizing equalizer among a Web site’s visitors.
You don’t need extra disposable income to read the latest news on “The New York Times” (at least today), or even on the BBC or in “The Telegraph” to get the U.K. perspective on world news and events. And that, as Martha Stewart would say, is a good thing.
What isn’t a good thing, according to ex-CEOs of some of these ad-supported-content sites, including Wenda Harris Millard and Jim Spanfeller, is that ad networks are part of the monetization solution that allows users free access to such great content. In their view, ad networks reduce the value of a publisher’s inventory by creating channel conflict and promoting hard ROI metrics over softer brand metrics. They propose that Web sites sell their own inventory directly to advertisers at high CPMs and leave the remainder of their inventory unsold.
Every Web site draws a mixed audience. Most publishers sell that audience like a bag of mixed nuts: an advertiser may only want cashews, but he has to buy the peanuts and almonds with them.
This works for a Web site that naturally draws a focused audience that meets the targets of an advertiser, such as Steve’s Digital Camera reviews. But what about a Web site that has such a heterogeneous audience that no clear category of commercial interest emerges to capture an advertiser, such as Reason Magazine, which draws an eclectic mix of well-educated readers.
Just as advertisers need a way to buy exactly the audiences they want, Web sites need a way to sell their ad space in a granular fashion to exactly the advertisers willing to pay the most. Ad exchanges offer an easy and compelling solution to buyers and sellers of ad space, allowing buyers to bid on context, behavior, geography, and a number of other factors that describe their ideal audience.
Don’t Blame Math, Blame Bad Math
To Millard’s pork-bellies manifesto, DoubleClick’s Michael Rubenstein responded, “We like to think of our publisher impressions as diamonds, not pork bellies,” meaning that relative differences in value among impressions are evaluated in a principled and fair way.
This isn’t entirely the case, because if buyers are evaluating online ad inventory based only on propensity to convert to a lead or sale, then they’re valuing only a component of the value. Many markets temporarily stabilize in this kind of regime of incorrect valuation, but in a frictionless economy buyers should emerge who understand the value and will gradually bid prices up to their fair value. So, if quality sites deliver brand value beyond the direct response metrics, can’t a clever ad network discover this and run brand campaigns through inventory on exchanges, and bid up the market?
Brand advertisers often measure brand value by the names of the sites a campaign ran on. If there’s a good story about why an audience that is likely to engage with their brand would be found on those sites, the advertiser is happy with the campaign.
This leads to advertisers and media buyers doing direct buys with Web sites for their brand campaigns. But if the industry could agree on a brand impact metric that was almost as easy to measure as CTR, then even brand advertisers could reap the efficiencies of exchanges, and publishers with great content could be rewarded for the brand impact of ads run on their pages, in addition to direct response generated.
Jerry Friedman is a famous Stanford professor of statistics. Early in his career he decided to spend his time writing computer programs to analyze data and experimenting with different techniques, rather than just spending time thinking abstractly and proving theorems.
This was a crazy decision at the time due to the primitive computers available even in university labs, but he realized that computers were getting faster all the time and that proving theorems was always going to be just as hard. It paid off. He ultimately invented many of the algorithms that underlie today’s successful ad networks and search engines.
Marketing and advertising will follow the same trends: computational approaches to finding the best-performing inventory for a campaign (or, dually, finding the best ways to monetize a Web site) using data on behavior, context, geography, demographics, and other factors will continually become easier and better.