Sarah Chubb has a unique "problem"...the sort more Web executives would like to have. The President of CondéNet, home of high-profile content sites such as Epicurious.com and Style.com, is running out of ad inventory for a very popular and lucrative category: healthy food products. Packaged goods manufacturers are clamoring for more and more health-conscious eyeballs, and they are willing to pay a premium CPM to nab them. But there are only so many contextually relevant places on Chubb's sites for advertisers to place these ads and still be confident of hitting the intended target.
Rather than investing in even more healthy-food editorial, Chubb is trying a more efficient solution to her "problem." She is starting to identify "healthy food" viewers as they enter her site and track where else on Epicurious they travel. It turns out that a lot of these health-conscious types also spend some of their time salivating over fat-drenched desserts. If you already know that this user favors healthy foods, then why not serve her an ad for a goody-two-shoes, feel-good-about-yourself low-fat snack, even if she is on a page that features a zillion-calorie slab of triple-fudge cake?
The point is a simple one, but few Web sites have been willing or able to realize it: the Internet allows publishers to sell advertising against users, not just against content. As Chubb says, "Sell them access to the people who access healthy eating [content], rather than just selling them the area of the site." Thus, without reallocating the editorial budget one penny or the site design a single pixel, CondéNet should be able to sell more of its existing ad inventory at higher CPMs. "The idea is that we start to segment our users based on their behaviors," says Chubb.
CondéNet, along with USAToday.com and Weather.com are beta-testing an emerging category of "audience management" site tools, software that moves a publisher beyond counting eyeballs and into the next stage of metrics, weighing eyeballs. "It is managing audience for profitability," says Dave Morgan, CEO of Tacoda Systems, which has designed the software engine behind these techniques. Through a system of user cookies and database mining, Morgan's Tacoda tools construct detailed profiles of how users interact with a site. It can even incorporate information about that user's behavior from offline subscription databases, or email and ad servers. From a single dashboard screen, the publisher then can analyze an audience for "how many people of what types we have seen in the last week, or month, and how many in profiles that are most valuable to us," says Morgan.
From Web Talk to Web Walk
The Web industry has always talked a good game about the potential for parsing and targeting groups of visitors according to their behavior, but the fact is that few publishers are even close to deploying these techniques. Weather.com is one of the most massive eyeball magnets online, with 10 million to 15 million unique users a month. Nevertheless, Joe Fiveash, senior vice president of product and business development admits, "Like most Web sites, we have only a rudimentary understanding of our audience." His eyeball bucket is so deep and diverse, that he will rely on Tacoda first and foremost to distinguish between the site's heavy users (15 or more visits a month) and more occasional ones. "A lot of people use us for a lot of different reasons," he says. "And we know that if we can talk to these people who use us for a particular reason we can get them to come to other parts."
Until Tacoda moves out of beta and into full deployment at these and other high-profile sites in the fall, the real profitability behind sophisticated audience management remains as theoretical as it has ever been. Remember ad targeting? That, too, was supposed to leverage the Web's unique power to market "one-to-one." Alas, both the creative and technical costs of nano-niche advertising proved too costly to implement, let alone recoup in higher CPMs. Worse, the advertising industry itself wasn't ready to buy such discrete slices of their market. Likewise, Tacoda is not cheap. The one-time software cost is $170,000, but there are subsequent renewal, service, and support fees. As well, Morgan says that many customers will need to invest approximately $60,000 in a data warehouse, extractors, and connectors.
Monitoring the ROI on an audience management approach such as this will be difficult. But Fiveash and Chubb anticipate Tacoda helping them develop new, premium-priced ad products as it also optimizes existing inventory, Morgan thinks his system could help tweak performance-based ad placements, perhaps enhancing CPC or CPA click-through rates by 5%, and that is something that would go directly to a publisher's bottom line. According to Forrester research director Charlene Li, the payoff for complex and costly audience management systems could be considerable. According to Li, "For media companies with significant traffic and—more importantly—significant advertising and/or subscription revenue potential, it can be more than worthwhile."
Audience management may have a deeper, subtler affect on the very culture of publishing, however. A system such as Tacoda can tell publishers exactly which portions of their sites are worth sustaining by giving very detailed metrics for the revenue produced by specific sections and even users. Sure, you can open a chat room or a message forum, maybe even a local news or sports area for your site, and just watch the eyeballs mount and the hang time climb. But are those users paying any attention to whatever ads, if any, you place there? If not, then are those loyal users ultimately valuable users? Do they move from the low CPM areas of your site onto the higher CPM areas where you do make money off of their attention span? Morgan promises that Tacoda can even determine how much money a site made off of every individual audience member by tracking her usage, the ads she saw, and the CPMs the publisher charged for them.
This could mean a radically different way for managers to view content and make business and editorial decisions. "It's a pretty dramatic difference in the way we'll understand the different clans of our audience," says Fiveash. Morgan recommends that a site be ruthless in tearing down even popular areas if they don't attract advertisers or push people into the money-making sections. Even in chat or message areas, notoriously poor ad venues, publishers need to invent ways of extracting value from those visitors, perhaps in the form of opt-in registration that lets a site market to them in other ways. Charlene Li wonders whether publishers really have the cajones for a Tacoda approach. "If a company determines that 80% of its profits, not revenue, come from only 5% of the audience," Li wonders, "are they ready to make the tough decisions that this implies? These include potentially foregoing growth and greater scale by focusing marketing, content, and sales primarily on that 5% of the audience."
And as with ad targeting, it remains an open question whether publishers and the ad industry both can accommodate the shift in metrics and thinking that precise audience management allows. Most agencies just want comparability, Chubb says, the ability to compare online reach and frequency in the same way they do offline. "On the Web you do not need to care about how many overall people [see an ad], you need to care about how many people I can deliver to your ad campaign." And, as Chubb points out, "That is hard enough for ad agencies to get their head around."