Behavioral Targeting: Then and Now

Behavioral targeting in online advertising has attracted plenty of attention lately. The past month or so has brought a series of new product and network announcements, along with news stories that highlight recent case studies.

Behavioral marketing is being touted for its ability to sell products or build brands in categories ranging from automotive to airlines and beverages. At the recent AD:TECH conference, industry analysts labeled it one of the next big things in online advertising.

For those who remember the failed attempts at behavioral targeting products and networks of 1999 and 2000 (InfoSeek, Engage, and DoubleClick were the most prominent), this flurry of attention has people wondering if, as Yogi Berra put it, this isn’t “déjà vu all over again.”

Why will behavioral targeting work now? Why didn’t it work then? I am asked, and have answered, those questions regularly. The differences between then and now.


The market wasn’t ready for behavioral targeting five years ago. Although a lot of money was moved around online advertising, it wasn’t well spent. At the time, the money was being spent by dot-com companies seeking better MediaMetrix numbers. All they cared about was buying quantities of pages. They couldn’t have cared less about the quality of people they were reaching.

That’s all changed. Traditional advertisers and marketers now call the online shots. They care greatly who sees their messages. They care about targeting. They care whether audiences they’re targeting are in the market for a car, are regular surfers of diet and fitness content, or regularly consume business travel content. They care about whether those audiences have seen their messages before and, if so, how they reacted.

Publishers care now, too. They care about maximizing audience and page value. They care about optimization. They care about inventory management. In 1999 and 2000, such issues were the furthest thing from their minds. Getting the insertion order was everything. No one cared much about who was in the audience. The only thing that mattered was the audience grew bigger every day.


In the dot-com run up, the idea of online one-to-one marketing was as overstated as it was under-delivered. Marketers were told it would soon be possible and practical to reach each and every one of their customers on the Internet through a simple combination of behavioral targeting, profiling, and database management. That this was never possible (or even practical, what large marketer wants to have a one-to-one relationship with 100 million customers?) didn’t seem to matter. Everyone drank the Kool-Aid.


Without much of a market for behavioral targeting, there was never much of a chance for technology vendors to fully develop products. Solutions were very basic and not very flexible. They weren’t scaled for the volume of audience, orders, or timeliness that’s essential today. Current solutions can, and must, support campaign delivery to tens of millions of unique audience members across billions of page views every week. They must update audience member profiles in real time to both be effective and meet market needs. This wasn’t possible five years ago. It’s a prerequisite today.

Today’s behavioral targeting technologies incorporate years of learning in the delivery of related applications and services in ad serving, Web analytics, and campaign management. They also have the benefit of technical standards that have been adopted by the online advertising industry and the consolidation of technology vendors over the past five years. It would be much more difficult to build and deploy behavioral targeting services for publishers if 1998’s eight major publisher-side ad-serving platforms still existed. Now, the industry is largely consolidated around two major vendors.

Business Model

The failure of 1999’s and 2000’s behavioral targeting solutions can be attributed as much to problems with business models as anything else. Problems over data ownership and privacy doomed the two most notable efforts. DoubleClick became inextricably enmeshed in a privacy controversy over its efforts to integrate offline Abacus data with its online consumer cookie database. Though its Boomerang product for marketers never went away, it took a back seat once the company abandoned its Intelligent Targeting initiative.

If privacy was DoubleClick’s bane, Engage’s was data ownership. Its Profilz product, in which publishers would cooperatively share behavioral profiles, was never broadly adopted, despite massive marketing efforts. Why? Engage owned the shared data. In the end, almost no companies enlisted in the service, excepting CMGI portfolio companies (Engage’s majority shareholder).

Although timing, hype, technology, and business model problems haven’t been fully resolved, the landscape has significantly changed since 1999. A real online ad market is looking for optimization services. Technology has developed out of a more mature market that’s learned from earlier efforts. And we have new business models than can survive, even thrive, in environments where consumers care about their privacy and publishers care about data ownership. Finally, the hype now is about real dollars earned by real publishers using real behavioral targeting for real advertisers.

What’s different between then and now? Plenty. The differences justify behavioral targeting’s future beyond flavor of the month.

Related reading