Making Web Site Behavior Adaptive
For most online marketers the goal of "Web 2.0" has been to create Web sites that attract and immerse visitors, prompting their maximum engagement with a rich variety of interactive content. In their single-minded pursuit of driving traffic, however, many marketers have failed to address a more fundamental goal: customizing and adapting their content to the personal needs of consumers, as Olivier Chaine, CEO of Magnify 360, explains below.
BI: How did Magnify 360's approach to what you call on-site behavioral targeting evolve?
Olivier Chaine: What motivated us to get involved in the behavioral arena was the fact that marketers online are now spending upwards of $18 billion to drive clicks. Yet all that money usually gets them if they're lucky is 2% to 3% conversions.
Behavioral targeting has addressed one facet of this dilemma. By identifying web visitors whose browsing or search behavior indicates particular interests advertisers can better target their media to drive higher quality clicks. But the focus has overwhelmingly remained on the advertising side. What's remained largely unrealized are developing ways of leveraging behavioral data to take the fullest advantage of users who do click through.
BI: How and why have developers of so-called Web 2.0 missed this?
Chaine: There are two default assumptions that have defined online marketing up to now as far as the role of Web sites. One is that the goal of behavioral targeting is to say people who are interested in product category X should land on page Y. The other is that your goal as a Web developer is to create the one best Web site. That, we believe, is a broken model. Actually in this regard online is way behind direct mail, where messages have been customized for decades.
BI: What about the role of analytics?
Chaine: When we look at the proliferation of all the data analysis and tools, there are so many processes to follow that most marketers are at a complete loss about how to leverage those in a practical way that will actually make a difference in terms of their user's experience and responsiveness. What we've done is build a self-learning platform to optimize user experience. What that means is, we free up the marketer by allowing them to focus on what their real job is, creative offers and messages.
BI: How does your concept of segmentation differ from conventional ‘off-site' BT?
Chaine: What we've found is that behaviorally segmented profiles based on general preferences or characteristics are useful, sure. But what gives you the biggest lift is knowing what visitors' priorities are at that moment -- not just that they're interested in a certain product category, but whether they' re actively shopping, just browsing or simply curious. The same person at different times can have different priorities as far as what information or messages would be most useful. Our system is geared to finding the behavioral cues that indicate those preferences. What we're saying is, if you want to really leverage behavioral data meaningfully, you have to think about Web site presentation in a far more adaptive way, not just one based on rigid rules.
BI: How does that work in practice? Any examples?
Chaine: We have a client, for instance, who is selling a cash register and sales management system. So say they had a prospect who had come to their site using the search query ‘POS,' which means point of sale. Our system would immediately begin profiling them based on what it could behaviorally model from the query itself. It would be clear they knew professional jargon, so their interests were professional. The profile would connect them to a micro-Web site stressing the benefits and advantages of the products. If they were to come during the day, we'd give them the essential information very quickly, because they'd likely be coming on at work. If they were coming at night we'd provide more detail.
Now say someone typed in the query ‘POS, inventory management.' The system would ascertain that they were very detail-oriented, inventory managers for instance. So, based just on that additional query term, the micro-site would have a very different look and feel. There would be more bullet points, fewer or no visuals and a stronger call to action.
BI: How about for a more consumer-oriented product?
Chaine: With consumers we'd be looking for behavioral patterns that give us clues about what types of information are going to be most critical to specific customer profiles in evaluating a product. For instance, some consumers are interested in product reviews and want objective comparative product data, others are focused on pricing and still others are looking at unique cutting-edge features. The system will not only identify profiles but target particular kinds of micro-sites based on them.
Another behaviorally based profile that may be very relevant comes out of mobile device use. If you know a visitor is an iPhone user, for instance, the system will, based on that data, make an initial hypothesis that they are likely to be interested in technical features as early adopters. So the ‘cool factor' would be important to presentation.
A BlackBerry user, on the other hand, will tend to be more utilitarian and want a very clear statement of value proposition in efficiency. Now of course each behavioral profile is constantly tested based on how people respond to and interact with offers, so it's adjusted continually adjusted over time.
BI: What are key goals for the next phase of on-site BT?
Chaine: We think that up until now most of the learning curve of marketers has been geared to finding out how to get clicks and build traffic. Marketers now are getting quite savvy about that part of the business. What they're still under-developed in is improving yield and conversions after the click. We think this year will be the year closing that gap becomes the priority.