I am not sure if Keith Piper and the crew he leads at Pretarget are fishermen, per se. But they sure do love to talk about "bass." Not the deep sounds that boom from my surround sound system, and not the
stringed instrument, but "bass" the fish.
Understanding the distinction among the possible meanings of "bass" is an idea that courses through almost all the promotional material the ad targeting company offers. It forms the core example Piper
cites when explaining how his company blends contextual and behavioral targeting. "Everything we do
comes down to the keyword," he tells me.
Pretarget was formed to expand upon the usual techniques employed by traditional behavioral
targeting. Piper, a veteran of the ad agency world, recalls his frustration as a media buyer with the
limitations of segmentation: "The segments we were being sold were irrelevant. 'Soccer moms' didn't apply to us in every campaign. We had a different target audience. We were trying to put a square peg in a round hole."
Coming off of a patent application he and a partner had made years ago, which mined content links
to extract keywords, Piper spun out what he sees as a different way of coming at the problem. "It
occurred to me that everyone buys keywords, and knows what keywords are and what they are
targeting with them." So his new company, Pretarget, starts with a list of keywords from a marketer
and then massages that list to conform to available data sets across a wide selection of data. "We use it to identify a variety of data sources and sets that we view as intent signals," he says. The data can
be gleaned from both contextual and behavioral in order to focus on users who are in-market with
intent. "The keyword is the seed to help identify segments and mash those components up into a
custom intensity segment," Piper says.
That is where the fish come in. "'Bass' can mean many things - a guitar, a beer, a shoe brand or fishing,"
he says. "If you look at just search data you don't know that, and if you look at page context you still
may not hit it. We look across contextual and behavioral sets. If we put them together, then what does
it tell us about a user? It gives us a better signal of what that user is looking for." It can also be applied
to many kinds of data sets to hone in on things like whether the user shopped for lures, or visited a bass
fishing page, or lives in a given region or shares content on fishing with others.
By then using the data to place the targeted ads in the contextually relevant places where that targeting
is most likely to show up, Piper says that the response has been substantially higher than when targeting
by context or behavioral alone: "Everyone has an assumption that content matters in making an
environment, and there is a general assumption that if you were to combine behavioral and contextual
then you can scale it better. We are hitting a user exhibiting a desired behavior in a contextually relevant
environment. The conversion rates really blew everyone away." Pretarget claims conversions of 20%,
compared to its own proprietary data on contextual (less than 15%) and behavioral alone (about 6%).
Calling it product "intentsity," Pretarget says its algorithms can also optimize on conversions to refine
the data sets it uses to re-weight towards those that perform better. In order to keep the system that
nimble, Pretarget relies on ad exchanges that offer real-time bidding.
Because you need to be quick to catch the right "bass," we guess.