The realization that Web site behavior is a pretty accurate indicator of service and product purchase intent is not new. As discussed in part one, behavioral targeting has been around for years, and data mining for longer. Most of the major behavioral targeting companies tell Web site advertisers and publishers alike that the real key to using this information more efficiently is to apply increasingly-sophisticated behavioral metrics to measuring and assessing what unique visitors do.
"One of the major improvements goes into the understanding of a lot of things that make up behavior -- everything someone does when he or she visits a site," says Brian Handly, CEO of Accipter Solutions, Inc. "Most of what is talked about is the behavior to a certain content area, or where and when someone made a purchase or filled out a form. It goes back to what type of action the person took.
"We can log that information and note that maybe this behavior happened twice in the last week, and maybe 20 times. This is not technology that knows who you are but where you have been," adds Handly, describing its VIBE product, or "Visitor Interest Behavioral Targeting Engine."
The initial installation of VIBE was with the online PostNewsweek Tech Media's Government Computer News, a magazine published for government buyers of information technology products and services, and Washington Technology, a magazine for government information technology systems integrators.
Accipiter says that PostNewsweek "understands that the most reliable indicators of a Web visitor's interests are how that individual interacts with content and behaves during online transactions. This is why behavioral targeting is the most effective way to create a meaningful connection with online customers and prospects."
In March, 24/7 Real Media introduced Insight ACT, an analytics technology that enables behavioral targeting through audience clustering. The company also introduced Open AdStream 5.7, the most current version of the company's ad management solution. The company says these two products provide its publishing customers with a fully-functional and totally integrated analytics, behavioral targeting and ad serving solution.
"In the last number of months, we have worked to build a seamless system, not only just behavioral marketing, but an inventory management module that accomplishes both analytics and behavioral advertising, while enabling publishers to control inventory," says Hills. "Now, we have both."
Insight ACT links Open AdStream's ad serving capability with Insight XE, which is 24/7 Real Media's proprietary content and commerce analytics system. Insight ACT offers behavioral segmentation and targeting in a three-step process of analysis, clustering and targeting of user segments. Insight ACT also provides inventory forecasting and control as well as performance reporting by behavioral segment.
Is contextual targeting better than observed behavior?
Claria Corp., of Redwood City, CA, views itself differently. CEO Scott Eagle says the "observed behavior" model, which some of his competitors use, can be just a baby step. "They don't know if that person just likes to read travel articles, or whether she is planning a trip domestically or overseas. Because they only have limited data, their model is still somewhat flawed."
Eagle prefers "contextual targeting," where specific behavior on many sites is tracked and aggregated. To him, a holy grail might be an interrelated series of discoveries that would work with user-inputted site registration information and surfing habits, to identify males 30 to 45 years old who not only read about travel, but make over $100,000 a year, check European art museum sites, and have typed in the keyword phrases "luxury hotels" and "Europe" into a search engine.
"Then, maybe you send him a Virgin Atlantic ad," Eagle says, referring to the British carrier with the contemporary passenger experience.
Claria's offering in this space is Feedback Research Database Analytics, a solution the company says "accesses the actual navigation behavior of over 43 million users across over 60 million domains every day." The secret sauce is information collected via the GAIN Network, a collection of free online behavior and advertising click-on measuring tools distributed to opt-in consumers. GAIN estimates a permission-based relationship with over 43 million users.
The data is then used by Claria to create comprehensive analyses of customer behavior in several categories, including which sites the user visited on the online session before clicking on the current site, on what site she shopped, what information, if any, can be gleaned by specific navigational patterns, the frequency of viewing and purchasing on measured sites, as well as comparative sales conversion levels.
As an example, Eagle cites a "big apparel company" that used information to design and target nine different "$5 off" ads aimed at women between the ages of 24 and 35, earn less than $100,000 a year, and through their surfing patterns were shown to be comfortable with buying online.
Is real-time helpful?
New York-based TACODA Systems says what sets it apart are the size of its customer base (18), the maturity of its technology (launched three years ago), and the ability of its technology to update profiles in real-time. The real-time capability means that an audience member's profile can be changed in milliseconds, according to specific browsing behaviors, such as which automotive content a user viewed or whether they filled out a questionnaire. Different ads can then be delivered to that person in the same session. CEO Dave Morgan cites one case in which a publisher delivered 50 percent more inventory in real time -- enough to cover the buyer's needs, and effective enough to win renewal later.
USAToday.com, iVillage, Tribune Interactive and Weather.com are among the more than 500 Web sites for which TACODA's Audience Management System handles segmentation and targeting. These publishers -- who collectively reach 70 percent of Internet visitors monthly -- have sold millions of dollars in incremental revenue through behavioral targeting and have learned what works best and what to avoid, Morgan says.
"Over the course of hundreds of campaigns, our customers and their advertisers have become the true industry experts," he says. "They know which data to use and what to leave out of a behavioral segment. They have tested thousands of combinations of frequency, recency and volume to create  segments that perform well in direct response and branding campaigns."
In addition, TACODA plans to launch an audience-centric ad network where ads will be sold in a Web-based auction. TACODA’s network will enable direct marketing and search advertisers to use behavioral targeting in combination with demographic data to reach defined audience segments of significant size across a vast network of quality publisher sites initially to include numerous sites that are already TACODA AMS customers.
TACODA’s network will launch offering text-based, pay-per-click ad units that will be sold in an open bidding auction process. The ads will be served to users wherever they are on Network sites enabling extremely fast delivery of highly relevant messages.
Role of relevance
Another behavioral targeting company, Bellevue, WA-based Revenue Science, approaches the space with the mantra "Reach, Quality and Trust. In order to satisfy the requirements of advertisers, publishers and consumers, we must focus in these three areas,” says Omar Tawakol, senior vice president of marketing.
Revenue Science is addressing reach, in part, through its full-service model. “Our revenue managers work closely with ad sales and ad operations executives at our publisher clients, such as Reuters.com and ESPN.com, as well as speaking directly with agencies and advertisers,” explains Tawakol. “By bringing both sides together to discuss how behavioral targeting works, and how to derive maximum benefit from integrating behavioral targeting into online campaigns, everyone gains a greater understanding, and we gain a wealth of knowledge on what both sides need to be successful, which informs the development of our services and technology.”
Revenue Science launched its Audience Search™ service last month with The Wall Street Journal Online at WSJ.com. Audience Search allows publishers to customize audiences at the campaign-level based on advertisers' RFPs. In addition, Audience Search provides a directional measure of audience quality through the Behavioral Relevance™ Score.
“This scientific verification of audience composition provides advertisers with the insight they need to invest in behavioral targeting as a mainstream form of online advertising,” Tawakol says.
Audience Search also addresses quality through a familiar interface that enables publishers to easily adjust the reach/relevance for every audience in real time, in order to achieve the optimal result for an advertiser’s campaign.
“What was once a complex, time-consuming process for publishers has become a simple task that can be done quickly and efficiently,” Tawakol says.
Finally, trust is quickly becoming the most important topic in online advertising. “We are committed to preserving the trust that exists between our publishers and their readers,” Tawakol says. “Revenue Science made the decision not to view or sell personally identifiable user information and to group all data on user behavior, providing complete anonymity. We only collect data within the confines of a single site, so users are not tracked across the Web, and we never install software on a user’s machine or use intrusive methods that disrupt the user experience.”
Different approaches, but one goal
The research methods between solutions may be different, but not so the goal. "It's all about the ability to understand and segment down audiences, by being able to see who is clicking and converting," Claria's Eagle says.
Or, as John Milton would say, "To help you find them."