–

Traditional segmentation

MM: If I understand that right, Trae, traditional consumer-segmentation efforts use census and other forms of compiled data such as credit histories and other forms of household data, creating a cluster or basically a neighborhood. Now that entails the presupposition that everyone in that neighborhood will buy a similar set of things; that consumers that share similar socioeconomic backgrounds and motivations will buy the same stuff. However, in reality, each household of a particular neighborhood often represents a huge difference of consumer appetites, criteria, and mind-styles. So the idea of Cohorting takes another approach: rather than working from the physical data (households of a neighborhood) to develop a data set, you now work backwards consumer appetites, criteria, and mind-styles, creating logical set that you call a cohort. This approach of cohorts or logical groupings of buyers makes neighborhoods like Swiss Cheese-where each hole represents a distinct set of buying criteria and, when group together, create a cohort.

Absolutely. In our approach to Cohorting, we incorporate attitudinal data (from surveys, interviews, focus groups) demographic data (from consumer or business databases) and behavioral data (from websites and syndicated sources). We certainly see big differences within neighborhoods along attitudinal and behavioral lines.

A lot of it has to do with the methodology we employ. For example, we may begin with broad and nationally-representative syndicated data like a Simmons or an MRI, enabling us to incorporate many of those things. Not just geographic, demographic data, but attitudes and behaviors across a broad swath of areas including brand usage.

Other times, where available, we’ll start with our client’s customer database-if they have a database-mapping those customers into syndicated or third-party data, or direct primary research.

By the way, since we’re talking about Targetbase and our approach…we define ourselves as data-agnostic and make it a point of differentiation. We’re not trying to sell data; it’s not part of what we provide. We don’t want bias our results and outcomes. It’s really more about the outcome, about the solution that we’re trying to deliver, and not about selling data.

Brand-portfolio mix optimization

MM: As you said, you can take one way of looking at the data in standard cluster analysis, using data from syndicated research or compiled data. Or you can segment along segmentation themes that reflect mindsets of various cohorts, supporting brand-portfolio mix optimization. So with a large portfolio of brands, segmentation themes enable you to say, “Okay, what is the optimum mix of brands that we should market, can market, have marketed, to this particular profile of behavioral and attitudinal data?”

Exactly. Like I said, there are certainly themes that crop up across clients, across verticals. But one example for a particular client is a segment that we have dubbed, “Maxxed Moms.” That particular group — some of the themes or points that really stick out are…They’re mothers and certainly time-crunched. Their time is important to them. However they also want to feel like they are pampering their family and taking care of their family in a very traditional sense.

Not only does that level of attitudinal behavior and understanding drive how we message to those members of the segment — the types of imagery, et cetera, that we use… but the cohorting analysis also tells us what products the client has a right to win with.

For example, if you’re going to that segment of Maxxed Mom, products that are geared toward time-saving and/or pampering your family. Home-cooked meals in this case might resonate very well with that segment.

So then we incorporate scientific test and control methodology to determine if — in fact — we can win with that particular segment, with that set of products.

–