Behavioral Targeting: Back to the Future?

There I was, sitting in the slick ’70s-style dining room at Manhattan’s Brasserie, listening to the young entrepreneur describe his vision for his company. Surrounded by power-breakfasters wearing smart suits, despite August’s drenching humidity, we talked about profile-based targeting across multiple sites, about building huge databases to store behavior-based user profiles, and about how the plan would allow advertisers to target relevant individuals like never before.

I felt like I was back in 1998. I almost asked the guy how much he was trying to raise in his next round of venture capital financing. Just another example of the way the industry seems to be going nowadays. Old (failed) ideas are being resurrected with new hope. They still sound great, but will they work this time around? Are they bad ideas, or was it just bad timing?

Remember DoubleClick’s Intelligent Targeting and Boomerang products? They were supposed to build profiles and target ads based on user behavior. Intelligent Targeting was abandoned in December 2001. Executives explained, given market conditions, advertisers weren’t willing to pay a premium for behavior-based targeting.

Then there was Engage, which touted itself as the originator of user profiles. Back in 2000, Engage said of its 70 million anonymous profiles:

These profiles contain rich, dynamic, anonymous behavior data that is processed based on: type of content viewed, the time spent viewing and the frequency and recency of visits to a particular interest category or site, all of which are based on 800 interest categories.

Engage, of course, filed for Chapter 11 bankruptcy in June. Its remaining assets were sold to JDA Software Group.

The Next Generation

Fast-forward to 2003. If you’re a media buyer or a publisher, you’ve probably heard terms such as “behavioral profiling” and “audience management” trotted out again. You hear them in pitches from publishers such as the New York Times Digital or The Wall Street Journal Online. Or in pitches from technology vendors such as Tacoda Systems, Revenue Science (formerly digiMine), or AlmondNet. They may remind you, as they do me, of days gone by. But this latest crop of technology companies — at least most of them — are much more realistic about the state of the market and the tools’ capabilities.

What these companies do, in simplest terms, is help publishers segment their audiences based on behavior on their sites. The Wall Street Journal Online, which is working with Revenue Science, offers advertisers segments such as “technology enthusiasts,” “car buffs,” “consumer techies,” “engaged investors,” “health nuts,” and “travel seekers.” The New York Times Digital, which developed a tool internally, offers “autos,” “pharmaceuticals,” “real estate,” and “tech/telecom.” It will soon offer “fashion,” “dining and wine,” and “home and garden.”

The idea is advertisers buying, say, the “travel seekers” segment would reach the same people they would if they were to buy in the travel section, but impressions are delivered elsewhere. Of course, this allows publishers to charge a higher CPM in less desirable parts of their sites, as the people are identified as desirable.

“I don’t believe that if the chairman of GM goes off to read the opera review that he’s any less valuable than when he’s in the auto section,” said Randy Kilgore, WSJ.com’s vice president of advertising, marketing, and sales.

That these technologies are emerging now and that publishers are starting to embrace them are good signs for the industry. They’re signs the most desirable site sections are beginning to sell out. They’re signs publishers are choosing different ways, other than simply adding smaller placements (i.e., clutter) to popular pages, to increase revenues.

The bad news is these tools may further complicate a problem that dogs publishers: inventory management. Key questions for publishers to ask themselves (and for media buyers to ask publishers) include:

  • How many times must people visit, say, the technology section, before they become “technology enthusiasts”?

  • How recent must those visits be, and when does someone get kicked out of a segment?
  • How do I make certain impressions are really there? If you’re selling an impression for a technology enthusiast visiting the world news section, make sure that impression isn’t already sold to an advertiser who wants to reach world news visitors.

There’s a balance, of course, between quality and quantity. The more narrowly you define segments, the better the response you’re likely to get. The downside is you’ll have less inventory to offer. Juggling inventory to ensure each impression gets the highest value possible is another challenge.

In the Field

Behavioral profiling seems to be working for some publishers. Belo Interactive, which publishes DallasNews.com and uses software from Tacoda, released two case studies this week touting results.

In one, a local auto dealer targeted banners to people who had visited the automotive section of DallasNews.com in the past 30 days. The response rate (i.e., CTR) came in at 7.7 percent, tremendously higher than the industry average. Of calls to the dealership, 44 percent were attributed to the DallasNews.com campaign, at a time when eight promotions were running in other media.

In the second case study, a vacation broker credited behavior-targeted ads for increasing vacation bookings by 50 percent and boosting revenue 38 percent.

These are some of the only hard numbers to emerge from behavior-targeting players so far, but media buyers are testing, watching, and waiting for more.

“I think it’s one of those things where we are still in the early stages,” said Nick Pahade of Beyond Interactive.

Jeff Lanctot, Avenue A’s vice president of media, says some of his clients are “dipping their toes in,” but it’s too early to tell what the results will be. “In general, I have a positive feeling about it,” he said, “because I don’t think I’m being sold a bill of goods by any means…. There’s a level of granularity that the publishers can provide that I think can be really helpful.”

A good sign for the future of this targeting technique is publishers seem to be taking a very realistic pricing approach. They charge a small premium in some cases, but not enough to discourage advertisers from trying something new. Revenue Science and AlmondNet have established revenue-sharing pricing models. They’re incentivized to make the technology work for publishers and advertisers. Two big clues we’re not reverting to 1999 or 2000.

What a relief!

Related reading

nurcin-erdogan-loeffler_wikipedia-definition-the-future_featured-image
pwc_experience-centre_hong-kong_featured-image
12919894_10154847711668475_3893080213398294388_n
kenneth_ning_emarsys_featured-image
<