Behavioral Targeting: Advertising Embraces the Internet

    September 27, 2004

Fresh off Ad Week and hearing the buzz about behavioral targeting, it seemed to make sense to put that term into perspective.

Behavioral marketing, despite its five year lifespan, had really just gotten off the ground in terms of product evolution at a time when the space – particularly online marketing and technology start-ups – was contracting. Now, as companies (re-)turn to data-driven marketing, there are some key techniques and principles that should be incorporated at the start of this renaissance to ensure solid results and sustainable growth.

To begin with, companies should look to their roots. Companies with the most success in online behavioral marketing are those that leverage their offline expertise in this area. Marketers have long relied on audience segmentation and clustering to find the right segments for their varied messages. Nevertheless, the online medium, despite inflated promises of accountability and measurability, has often relied on broad strokes with a dull knife to optimize campaigns. Without behavioral targeting – some means of differentiating an audience based on how they act – a typical “optimization” routine finds advertisers looking at which ad has the highest average response rate per site, and then shifting all impressions toward those handful of best-performers overall.

There are two main problems with this approach. The first is that, in the worst case, the revised targeting relies on click-through as a measure for “performance”, a dubious metric at best (anyone could generate a lot of clicks for a credit card offer with the message “$1,000 free with any application, approved or not!” but very few legitimate prospective cardholders). Simple conversion rate, measured either on a click-through or view-through basis, is the more recent proxy but remains one which is still too removed from bottom-line results to be predictive as a metric for success.

Secondly, decisions made in the aggregate, without taking into account variation in audience response, tend to obscure a tremendous amount of valuable information. It reminds one of the old saw about having one hand in a bonfire and the other in a bucket of ice water and being “comfortable, on average”. Rather than having to rely on the single message that weakly appeals to everyone, behavioral targeting offers the promise of finding the dozens of messages that strongly appeal to each segment of your audience and making sure to match them.

Which again conforms to the offline world in which sophisticated marketers constantly expand the scope of their offerings with a new offer for every demographic, interest group, purchasing behavior, etc. They look to blanket the universe of prospects with as many offers as necessary to maximize a complex function of conversion rate, purchase price, customer lifetime value, and more. In short, they engage aggressively in audience segmentation and behavioral targeting.

Behavioral targeting online can now use these tried-and-true offline practices and leverage them in a new channel that offers its own set of unique strengths. Developing a successful data-driven campaign requires a few key elements:

Quantify all goals. It’s critical to assign values to the various events that can be tracked. While a retail store generally has to rely on register data and some anecdotal information about how people physically move through aisles, online behaviors can be much more carefully observed. But they have to mean something. Start somewhere: maybe site traffic is considered a good indicator of popularity, and can translate into media sales on the other side. Let’s value a visit or interaction with a rich media brand message as a “1”. Starting a registration or submitting an email address is worth 2. Purchases can scale according to total order size or modeled customer lifetime value. Yes, it is often hard to quantify certain behaviors, but deciding to not make a decision is a decision in itself. Setting up a consistent benchmark to evaluate performance – flawed or not – is a first step to beating that benchmark.

Integrate channels. Online media tends to be an extremely cost-effective channel to “prime the pump” with exposures to numerous prospects. How do they respond to offers? How, based on the learnings of actual customers, can those results be extrapolated to the prospects that look just like them but haven’t yet purchased? How can you best integrate direct response and brand – to prove that the right sequence is one relatively expensive (to develop and serve) rich media creative per week, followed by a rotation of increasingly sweetened direct response messages with hard calls to action? Any vehicle in isolation could sputter, but together, they can move people down and through the classic purchase funnel faster than another channel. Campaigns managed in silos – where each business unit ignores the efforts of all others – are necessarily inefficient. Companies that integrate efforts in acquisition, cross-sell, retention, brand and direct response across all product lines find tremendous improvements in overall performance.

Keep a consistent picture of the user. Now that you’re using portfolio management to deal with your inventory, use the same broad insights across your audience. Knowing that someone expressed interest by clicking on an ad but abandoning the registration process suggests now may be the time for “cancel/save” offer that allows them to complete registration at a discount rate. Or for a dormant customer, feel free to talk to them through online messaging about your new wider seats, DirectTV on all flights, and hopefully soon, in-flight wireless access features through email, affiliate marketing, or the homepage of your site. Understanding where people are in relation to key activities the marketer is trying to promote and rigorously testing strategies on how to move them in the right direction is crucial.

Summary. At the end of the day, it’s a straightforward relationship: the companies that rely the most on the channel as a source of purchases or registrations or email submissions tend to be the most innovative and motivated. The channel has proven an inexpensive test bed at $1.00 per thousand for how a direct mail piece will perform at $2.00 each (2000x the cost). This next phase in marketing will be data-driven. While the banner ad or the e-commerce web site will not, as the TV commercial folks tell us, “make us cry” by being able to evoke emotion, it can tell us much about how to get there and how to turn the evocation into a transaction.

Joe Zawadzki is the President and Founder of Poindexter Systems, based in New York, New York. Poindexter Systems is a leading provider of performance optimization technologies that enable online marketers to maximize the performance and profitability of their Internet-based advertising and media campaigns. Joe can be reached at