Slicing and Dicing a Captive Audience

Since the days of the broadside, publishers have tried to gain a better understanding of the people reading, commenting on, or interacting with their content. The ability to track user behavior over time not only can help Web publishers learn more about their visitors, it can create revenue from otherwise undervalued content. But publishers remain cautious, especially as they become more aware of the headaches behavioral targeting can bring.

Cheerleaders for behavioral marketing techniques say the latest data tracking and audience segmenting technologies allow publishers to truly define their audiences. Finally, they say, publishers have the potential to sell all their inventory at a premium. Now that publishers can individually track and serve ads to dieting divas or celebrity gossip junkies, advertisers won't clamor only for high-demand contextual placements, but for any page their target market visits.

But this audience-slicing sword is double-edged. Sometimes advertisers want to reach more people than single sites can offer. To create higher volume for advertisers, sites can join networks, so data can be shared across multiple sites rather than within just one. But this could dilute the unique value of each publisher's content, as well as create privacy concerns. In addition, advertisers want more standardization of audience segments to streamline media buys. This, too, could weaken the unique worth of each publisher's audience. Plus, behavioral-targeted ads can wreak havoc on inventory management, compounding an already unpredictable problem.

Harvesting "Left-handed Rose Growers" Despite these arguable pitfalls, publishers like Dow Jones Online can take what it already knows about its audiences and refine audience segments based on page views. The publisher has determined that 99 percent of Wall Street Journal Online ( readers are college-educated, and have household incomes of around $215,000 on average. This site's audience can be divided among pre-defined segments such as "Financial Elite" and "College and MBA Students," or can be broken down according to advertiser objectives.

"Say someone is selling a PDA [personal digital assistant] and is looking for early adopters of technology," describes Mike Henry, vice president of sales and marketing at Dow Jones Online. The site can isolate all the people who visited the Online Journal's Gadgets column during a given period, "and those people get put into the 'Gadget People Basket'," Henry explains.

"We're usually trying to package our content, not our users," Henry adds. has been using Revenue Science's behavioral targeting software for about two years. At first glance, this seems like nothing more than contextual targeting. The difference, of course, is that with behavioral targeting, the ad can be served anywhere the reader goes on the site. So, if a targeted 'Gadget Person' visits an area of the site that garners typically low cost-per-thousand (CPM) rates, the publisher can charge more for that inventory than in the past.

"We've been intrigued at the idea of a specific audience that's being sold at a lower CPM that can be monetized better," comments John Kosner, senior vice president of new media at Despite the obvious appeal of rate hikes, the sports publisher has yet to fully embrace behavioral targeting. "Our take is it's early in the game," Kosner says, adding, "While there is advertiser interest, we've had no problem booking inventory through normal placements."

Jim Spanfeller, president and CEO of, isn't convinced that behavioral targeting is right for the financial site. "Our site tends to be homogeneous. Business decision makers are our target. [Behavioral targeting] is better for a larger, multifaceted site." has experimented with behavioral targeting using Tacoda Systems' software and its own in-house technology.

"In our case," explains Peter Naylor, senior vice president of sales at iVillage, "I've got a nice astrology destination, but there aren't endemic advertisers for astrology." The publisher employs Tacoda's technology to enable behavioral targeting on iVillage and, an iVillage property. Audience segments are determined according to the types of content users view, newsletters they've registered for, and information gleaned through online quizzes.

The publisher puts its users through a slightly more stringent approval process than some sites offering similar audience categories. In order to become an iVillage "Fitness Guru," a user must have visited the site's Diet and Fitness channel three or more times in 30 days; other sites may require just one or two visits to a similar section.

iVillage developed a set of audience segments based on a hair-coloring quiz for a beauty advertiser. Based on their responses to a quiz that associated hair color with self-expression, users were divided into five segments like "Life of the Party" and "Classic Communicator." Now, no matter what section of iVillage they visit, quiz responders are served ads for the hair-color brand that correlates with their quiz-related segment.

"For some vertical areas, there's only so much traffic certain areas will generate," says's Spanfeller, referring to the small number of people who fit into narrow segments like "left-handed rose growers." Scale and Standards: The Catch-22

Lack of scale is one of the biggest barriers to advertiser adoption of behavioral targeting. Some believe network models are better than having to purchase media on a site-by-site basis. While networks may spur more advertiser usage of behavioral targeting, which would help raise publisher CPMs, not all publishers are keen on the idea.

"That's an angle that commoditizes what we're doing editorially," contends iVillage's Naylor. "It's like saying a page in The New Yorker should be the same as a page in Parade magazine because they all came off the printing press."

"The unique selling proposition is present with that Web site, but those advertisers also have to worry about reach and frequency," suggests Jarvis Coffin, president and CEO of ad network Burst! Media. Burst! began testing Tacoda's technology on its network in March, and plans to create audience segments according to advertiser needs. Coffin expects Burst's segments  perhaps "Financial Risk Takers" or "Sideline Sports Fans"  to be based on its 477 contextual categories and subcategories.

"Scale is a problem in the uniformity of customer segments. [The lack of scale] is very frustrating," asserts Jason Krebs, vice president of sales and marketing at The New York Times on the Web. The fact that standard audience segments don't exist, is an impediment to the growth of behavioral targeting. Even if they must purchase media from each site individually, advertisers insist that standardizing segments to follow the same sets of criteria across all sites would help spur that growth.

There's a tug-of-war between the need to boost advertiser adoption to create higher ad revenue and the desire to preserve each publisher's unique voice. Michael Zimbalist, president of the Online Publishers Association, compares Yahoo! Health section visitors to WebMD's audience, and cautions, "It's very, very difficult to try to standardize across the full universe of the Web."

"While their brands are important, at the end of the day [publishers have to] sell real estate," concludes Rich Person, chairman and CEO of Poindexter Systems, a provider of ad-serving and optimization technology that enables advertisers to place bids on specific audience segments.

Denise Garcia, research director and principal analyst at GartnerG2 recommends a compromise: "Publishers will want to reserve a certain amount of inventory for ad networks, but put restrictions on ad sizes or placements." In this way, advertisers that buy directly through the publisher "will differentiate themselves." Managing inventory alone, much less off-loading run-of-site placements, is a dilemma for many publishers. Behavioral targeting adds to the confusion since it requires predicting how many people within a certain audience segment will visit a site.

"We look at how many we have available and cut it in half. There's no way of judging the impact," notes Randy Kilgore, senior vice president of advertising for Dow Jones Online. "It cramps our ability to monetize."

Behavioral Targeting Isn't Just for Ads Anymore The reality is that publishers can apply what they learn about their site visitors to what's attracting them to the site in the first place  the content. Behavioral data can be used to inform editorial, content placement, and day-parting decisions.

"With behavioral, you could be targeting more relevant content to people on the fly  dynamically, automatically," declares Scott Eagle, senior vice president of marketing at Claria. The adware firm plans to introduce a product that will allow publishers to serve customized content to users based on behavioral data.

While publishers analyze site traffic and other intelligence to make editorial decisions, few currently take advantage of behavioral information to enhance site content. Employing behavioral data to change or target content isn't in's plans at this point. However, "Personalization will be a major holy grail of the Web, especially for content sites," Forbes' Spanfeller notes.

Using behavioral data to target consumers, says iVillage's Naylor, "is part of the recipe for success, but there are a lot of other ingredients. Some are very simple: we look at where people click; we see what the message boards are saying. All those things together direct [editorial content], including behavior, but we're not leaning on just one thing."

Next story loading loading..