As we encountered in last week's interview with 7 Billion People, understanding the psychology of purchase behaviors and segmenting online audiences accordingly is one of the growth areas for behavioral methods. But where is the best place to capture the behaviors that are most relevant to your company's goals?
This week we explore yet another approach that tries to apply academic research to consumers. The two-year-old Decision Tactics method actually puts consumers into a product-specific decision matrix to discern the major purchasing segments for an item and each group's key purchase drivers. Co-founder Shlomo Ron explains how it worked recently for Gateway Computers.
Behavioral Insider: How is your approach different from other consumer segmentation?
Shlomo Ron: It was developed by professor Alex Mintz, a leading expert in decision modeling. His computerized decision process tracing method combines sophisticated data mining with cognitive psychological analysis. Initially it was an experimental research tool for optimizing decision-making processes of high-ranking US Air Force officers.
A lot of BT companies analyze behaviors on a Web site by tracking visitors' keywords and interaction with pages. We come from decision science. We identify the decision profiles or segments that exist. We developed a Web-based tool that sits on our server, called Live Behavioral Segmentation (LBS). Our method comes before you actually make the purchase decision. We are not looking at purchase history; we are observing behavior as they are making decisions.
We actually complement the other methodologies. We track how consumers aggregate product information into a decision. The output is customer segments based on their decision behavior and their corresponding decision drivers.
BI: What kinds of segments are we talking about?
Ron: We did an implementation for Gateway Computers to identify customer segments of notebook shoppers. We focus only on product pages. We had a link to compare notebooks at a glance with links to three price ranges. You land on our tool that looks like a matrix and replicates the product info you see in the comparison chart.
But the typical chart bombards you with information about features and speed. We show the same matrix but all the cells are invisible. In order to find what you are interested in [storage, memory, etc.] you must click on a cell to see specific information. That way we can then track the info that really matters to you in making a decision.
BI: So you are actually determining their priorities through their direct interactions?
Ron: Right. To the user it looks like a merchandising unit. We have a decision algorithm in the back that tracks the different decision trees that users are going through. For example, one segment of bargain shoppers go directly to the cheapest product and just focus on those features. Then you might have the inverse picture, which is the power shopper, but with more expensive models. Then you might find what we call the ‘single feature segment' interested in comparing memory across models. So we aggregate those behaviors. Unlike other customer segmentation methodologies that have preconceived segments, we have no idea what we are going to find. It is all organically learning the segments as we go.
BI: But at some point you have to determine what behaviors constitute a viable segment.
Ron: Sure. Our tool takes 24 to 48 hours to have a representative sample size. Then we determine based on specific thresholds of specific behaviors that it is a significant and distinct segment. We have the algorithms that tell us we have a consistent and distinct behavior pattern and it is not random or blurry. Once we have the segments, we also know what are the decision drivers and the decision disregards.
BI: So with the Gateway instance, how many segments would you come out with?
Ron: It usually runs into five to six. We found over time that each product line has undercurrent segments that need to be revealed. Then you pretty much know how to talk to each segment, their corresponding decision drivers. Let's say we know ‘bargain shoppers' are converting very well, and we know that three drivers trigger their decision. All they care about is memory, storage and free shipping, for example. Once we know that and we know that they disregard maybe processor speed, we have their specific DNA and know how to talk to these people. Then you can implement that to the Web site and to outbound marketing campaigns.
BI: Outside of onsite segmentation, what are the other ways this knowledge can be applied more broadly?
Ron: The other implementation is for research professionals. We helped a research company figure out the impact of brand and price on the purchase behavior of users buying digital cameras. We use online panelists that fill out a survey. Then they are directed to our LBS tool, so we can get more granular information not only about the specific decision profiles of different consumers but also their demographics, psychographics, etc.
We found that most people actually are interested in the cheapest price range for digital cameras and within this price range we found that they lean towards the premium brands. We also found that the top decision drivers for digital cameras are across price ranges, megapixels, formats and digital zoom. Then you have the lingua franca you need to include in every campaign.
BI: It seems that this approach somehow could layer onto existing BT systems like the ad networks to understand segments better.
Ron: We are in talks with some of the leading networks. Once we can create an intelligence that over time can inform advertisers about specific customer segments and decision behaviors, they can feed that information into their ad management services and be able to talk to those audiences. We believe that each product category has its own segments or decision profiles that need to be revealed. It is like turning the light on in a dark room. You just need to reveal those decision behaviors, and then convert those languages into action.