Who Are Those Guys?

Butch’s persistent question to Sundance in “Butch Cassidy and the Sundance Kid” about the lawmen hunting them is the mantra of e-commerce marketers. Understanding who is shopping, kicking tires, masquerading as a 14-year-old cheerleader and filling shopping carts is critical to efficiently building sites, crafting e-commerce tools and buying online media.

Understanding our customers on a granular, one-to-one basis is the critical variable for developing product, pricing and promotional strategies online. So far, we have little data and are stuck with the limitations of our own experience, techno-bias, race and class prejudice skewed even further from reality by the self-serving research blurb of the week or software fad of the moment.

In this context, Engage Technologies steps forward with the promise of anonymous profiling, an oxymoron if ever I’ve heard one. These guys audaciously think that aggregating millions of surfing sessions and searching for patterns among them will yield meaningful profiles of imminent buyers, likely prospects and profitable segments in ways that can meaningfully lift response and sales from Web sites and online ads.

The logic chain begins with some direct marketing fundamentals. Engage’s CTO and co-founder Daniel Jaye sees a universe of customers unwilling or unable to part with reliable or projectable data sets. Citing the needs of privacy, ego and American individualism, Engage eliminates the prospect of widespread data collection on the Internet from the get-go.

Instead, it starts with the premise that millions of people surf the Web, leave a trail, accept cookies and won’t object to collecting and analyzing these patterns anonymously. Drawing on the DM catechism that behavior is always more predictable than stated intention, and citing that almost 95 percent of surfers accept cookies, Engage reasons we can work with the data readily available and aggregate traffic patterns in statistically valid ways to predict the interests of visitors and likely customers.

Putting the Data to Work

Engage starts with the practical notion that we ought to use what we’ve got, recognizing that it is not perfect or even as close as offline database techniques. But, it says, the result could be a valuable incremental targeting tool that can increase the efficiency of our ad and design spending.

Dan Jaye is no braggart. His candor about the product in its 1.0 configuration is endearing. Though you are still left wondering whether this process is better than any savvy marketer’s intuition. The product promises an incremental targeting value over content-based buying at costs touted to be “significantly less than per-impression prices.”

For instance, Engage says don’t buy Golf.com to reach golf enthusiasts at $30 per thousand. Instead, it claims, better target golfers by buying a series of sites whose visitor surfing patterns indicate the presence of serious golfers at a much lower price.

There is a promise of “greater or equal response rates” baked into the service when a client buys a minimum of 100,000 impressions per week.

Consider the Engage logic chain. It begins with the well-documented understanding that geographic concentrations of demographically and like-minded people exist and that clustering and database analytics of many stripes rely on the notion that “birds of a feather flock together.”

For Jaye, the logical extension of this database dogma is “birds of a feather surf together” and that “surfing behavior can be a surrogate for brand awareness, consideration and purchase intent,” especially if repeated patterns can be found and validated by comparison against a statistically significant control group. That’s the net result of the algorithms now in beta test.

And while the company’s PR flacks claim to have 30 million profiles of surfing patterns, Jaye says Engage benchmarks patterns against a seed database of 100,000 well-scrubbed, deduped profiles which can be assessed and reported by consumer interest, geography and demographics. He’s not sure how many customer profiles the company is willing to promise eager marketers.

Working out the Kinks

But keep your expectations in check. The algorithms are new and being “tweaked,” the geography is inferred and doesn’t really finesse AOL’s masking problem. Plus, the demographics are self-reported.

If this doesn’t raise an eyebrow, understand that the profiling engine is linked to Engage’s ad serving business, so skeptics and hard-core databasers have reason to doubt the intellectual purity of their process.

The whole act turns on the credibility of the profiles, which are based on observed surfing patterns and “limited” transactional data.

Some of the key variables that shape the predictive model include how often a consumer visits a site, the number of occasions consumers visit a particular page, how long they look at each page, the recency of visits and the number of repetitive visits within a given time frame.

Calling Engage analogous to the offline RFM formulation, Jaye says profilers can discern groups of consumers with similar interests in site content, class of sites, or class of pages.

Engage’s confidence in the predictability of data is a function of number, repetition and consistency of visits. Jaye argues that by scanning millions of traffic patterns Engage can identify patterns of visits which “are indicative of age, gender, the presence of children in a household” and other potent variables.

As with other profiling tools, Engage cautions that the finer the filter, the lower the yield. It is an idea that has always struck me as a pre-emptive apology for the built-in limits of the tool. Jaye recommends this tool be used by clients looking to “deliver specific response-driven messages” or by agencies looking to reach “fairly targeted, large-scale audiences.”

Engage intrigues and bedevils me. I have a hard time believing my wacky, nonlinear surfing reveals anything about me other than a cheapskate’s unwillingness to pay for dirty pictures. Yet as a veteran database marketer, I know that using surrogate data points and sifting through terabytes of data have yielded up all kinds of insights for my clients. In the end, I want to see the math.

Related Posts