Soaring customer acquisition costs are devouring many online retailers’ margins. On-site behavioral targeting can be a powerful conversion driver—and deliver increased profitability.

By Tom Dellner

The term “behavioral targeting” carries a variety of connotations-most of them negative. For some, it conjures Orwellian images of some enigmatic Big Brother watching your every move on the Internet.
Into whose hands will this data fall? To what uses will it be put?

To some skeptics, it sounds like fuzzy science. Are we to believe that a few clicks of a mouse will provide us enough information to develop a reliable profile of a customer, giving us actionable data that will help us more effectively merchandize our products and dramatically improve conversion rates?

For almost all of us, the phrase leads to a certain amount of confusion. The term is so broadly used and frequently misapplied that few of us have any more than a vague understanding of what it really means.

WHAT IT IS-AND ISN’T
There are two categories of behavioral targeting. Network behavioral targeting-or behavioral advertising-is used by ad networks that gather data on a user’s site-to-site browsing behavior to build a demographic profile of the user in order to serve him or her more relevant ads on the sites within the network. It’s this sort of behavioral targeting that has some consumer groups and privacy advocates concerned.

The subject of this article, on-site behavioral targeting-sometimes called behavioral merchandizing-is quite different. It uses data-how a user navigated to the site and the various browsing behaviors the user displays once on the site-to serve a better and more relevant on-site experience, with content, offers, promotions, ads and product recommendations personalized for the user. It doesn’t rely on site-to-site browsing behavior and raises far fewer privacy concerns. In fact, it’s little different than what e-commerce professionals have done for years: study consumer behavior on their websites in order to refine the site and improve their users’ online experience. It’s just that behavioral targeting is automated, scalable, more accurate and more personalized.

So where does this leave site search or “searchandizing”? These solutions help a retailer deliver more relevant results with its site search function. It can also be implemented so that search terms trigger any of a variety of merchandizing rules, improving conversion rates and increasing cross-sell and upsell opportunities.

While an effective solution for many, site search is business rules-based (set up manually by the online retailer and/or solution vendor), whereas behavioral targeting solutions are “self-learning.” Site search results are predominantly pre-determined; they don’t change based on a user’s behavior-they vary only with the search term. However, in many cases, the two types of solutions can be used together-the behaviorally targeted solutions can help enhance and automate the search functionality.

MULTIPLE SOLUTIONS: ONE GOAL
On-site behavioral targeting solutions take many forms, but the mission is the same: drive conversions and revenue. “Most of an online retailer’s margins are spent in trying to drive traffic,” explains Meyar Sheik, CEO of Certona, the San Diego-based creator of the popular Resonance optimization and personalization platform. “There’s been quite a bit of innovation on the traffic side. But over the same period of time, conversion rates have continued to hover at 2.5 to 3 percent. The other 97 percent needs to be engaged within a few clicks, with the right content put in front of them. Only then can we begin to maximize conversion events and revenue.”

Rolf Elmer, CEO of the Swedish company Avail Intelligence-a leading provider of behavioral merchandizing solutions in Europe, now eyeing the American market-agrees. “The cost of acquiring a new customer today is approximately $6 to $8; e-tailers are really feeling the pain. Behavioral merchandizing has gone from ‘nice to have’ to ‘must have’ in a short period of time.”

Most of these solutions work by collecting vast amounts of data from each user visit: the time of day, day of week, whether the user came in via a particular search engine or pay-per-click campaign and even the specific operating system used. Every click is recorded: page views, searches and transactions. Tendencies and affinities are identified. More relevant content-ads, offers, promotions, articles, white papers and product suggestions-is served based on: a user’s past transactions and browsing behavior; the accumulated data of other users who have engaged in similar transactions and behavior; and-in the case of real-time solutions-the user’s current click stream. “We profile every visitor click by click in real time,” says Sheik. “After a click, the system can search 2 or 3 million product or content choices and choose and serve the right ones in 200 milliseconds, round-trip.”

“Self-learning” on-site behavioral targeting products get “smarter” with additional traffic and accumulated data. That is, they are in a state of constant refinement, becoming more accurate on a daily basis.
There are solutions designed for virtually any business model. Some-like Certona’s Resonance and those offered by Avail Intelligence-are ideal for online retailers with complex product catalogs and tens of thousands of SKUs (or more), lots of product turnover and high traffic volume. Others, like Omniture’s Test & Target segment-based tool, are better suited for mid-level sites with less traffic. Omniture’s TouchClarity product is a real-time solution that reads visitor characteristics to serve the most relevant of one of 10 or more on-site promotional ads to direct that visitor down the correct vertical of the site. For example, a banking institution could implement the system to profile visitors-in real time-to more accurately serve one customer credit card promotions, while showing the next visitor to the site promotions related to home loans.

The MarketMaestro behavioral analytics solution from 7 Billion People (headquartered in Austin, Texas) takes a very interesting and entirely different approach. Instead of focusing on product or offer relevance, 7 Billion People attempts to personalize the website experience on a communications level.
“Humans have been doing face-to-face commerce for thousands of years,” says CEO Mark Nagaitis. “If you walk into a store, 50 percent of the time you’ll walk out with a product. The 2.5 percent close rates online are just not acceptable.

“Utilizing our team of behavioral scientists, we have developed a neural network based on linguistics. We gather data-or clues-from clicks, navigations, searches, etc., and feed it through the network. It scores the clues in a way that enables us to create a ‘portrait’ of each visitor based on their motivations for this online session, how they want information presented to them and how they tend to make decisions.” MarketMaestro’s behavioral analytics highlights critical areas of the website where there is “dissonance,” or a misalignment of communication preferences between the site and the visitor-and makes suggestions regarding how to fix these key disconnects. Your site essentially develops the ability-like a good sales person-to customize its pitch to the particular visitor.

THE NUMBERS DON’T LIE
What can you expect in terms of results? “We typically see up to 15 percent in revenue lift, depending on how well-merchandized the site was to begin with and how the targeted content is integrated into the site design,” says Certona’s Sheik.

Personal Creations, a leading provider of personalized gifts, uses Certona’s Resonance and attributes 20 percent of its online revenue to the solution. “The average order value from those who rely on the automated recommendations is more than 60 percent greater than those who don’t,” says Nancy White, Personal Creations’ director of e-commerce.

Avail’s Elmer says that conversion rate increases of between 10 and 20 percent are common, as is incremental revenue lift of 20 to 30 percent.

Even better, many of the solution providers are compensated as a percentage of incremental lift. In other words, if they don’t perform, you don’t pay. Other vendors are often willing to run a cost-free trial to demonstrate the results you may reasonably expect.

IMPLEMENTATION
When choosing a vendor, select one with a solution that’s well-suited to your business’s vertical (ask whether the vendor has done business with any companies in your-or a comparable-vertical and whether they were able to provide revenue lift). Consider the type of content that you’d like to personalize (e.g., ads, promotions, product recommendations, white papers or other content) and make sure the vendor’s solution is capable of delivering. (Some specialize in ads and promotions while others focus on product recommendations; a few offer both.) Determine whether the product delivers personalized content in real time or whether there is a lag (of several seconds, a day or even a week, in some cases). If there is a lag, the solution will impact only similar subsequent visitors or the current user on a return visit. Understand the potential impact of user action (ad blocking, cookie deletion or cookie blocking) on the solution’s effectiveness.

You’ll also want to know how long the implementation process is and how much effort you’ll need to devote. Some of the self-learning solutions may have to run for as long as three or four weeks before they are effective. If so, ask whether they have a rules-based system to utilize until the self-learning tool is ready to go.

Finally, find out if the solution has implications to other channels. Some of the most robust products enable you to personalize confirmation e-mails or your e-mail newsletter. They can even be synchronized with your call center to enable agents to make more relevant cross-sell and upsell suggestions. You may be able to use the system to enhance your paid search or direct-mail campaigns, as well.

Once you’ve chosen a solution, follow a few basic guidelines: Involve your site design team. Even the most targeted content will fail to yield results if it’s not integrated into the overall site design and featured prominently “above the fold,” with photos if possible. And if a self-learning solution permits you to override it with merchandizing rules, don’t overdue it. These solutions perform best when allowed to run with minimal restriction. In addition, consumers are very perceptive and can often recognize when recommendations and results are being manipulated to “sell them something.”