Do We Need Geodemographic Clusters Anymore?
IMAGE SOURCE: PERSONICX.COM
They gave us segmentation definitions with names like Beltway Boomers, Apple Pie Families and Golden Year Guardians – and for 30-40 years, it was the best game in town for helping to paint a richer picture of who we should target, and what made them tick.
I’m talking, of course, about cluster-based segments (or referred by some as “geodemographic systems”) from products such as Prizm (Nielsen), PersonicX (Acxiom) and Mosaic (Experian). By looking at how customer households in a data file match the geo-definition clusters defined by these systems at a block level, zip/SCF (sectional center facility), or custom definition, marketers could get a better idea of the demographic, lifestyle, purchase propensity and media habits of the segments within those geographies.
The Achilles heel of these products is that they are based on householdlevel data. With their broad-stroke definitions, and propensity to oftentimes be suspect in areas that lacked homogeneity, it now raises the question: Where do segmentation clusters fit in in the new world order of micro-audience targeting?
After all, marketers now have what seems like an infinite quantity of data at the individual level: browsing/research behavior, content consumption, transaction history, service inquiries, social network activity and the like – which can be queried with agility and greater precision. And survey-based profiling provided by companies such as MRI, Simmons or Resonate, allow us to query by product to back into hundreds if not thousands of variables from which to paint a picture for insights, segmentation, persona development, or audience definition.
So, are geodemographic systems ready to go the way of the Walkman?
Not so fast. Here are but a few practical applications:
Your company has lagged in establishing an integrated data infrastructure – so your customer file is about the only true asset you currently have. Matching it to Prizm, PersonicX or Mosaic is a viable way to determine if your customer base is actually who you think it is ... and to jumpstart your segmentation to provide strategic insights that can be acted upon. Over time, as you accumulate more data assets, you can challenge, refine or replace these definitions to create new, more actionable designations.
There are products and services that can benefit from, or are vulnerable to, what I’ll call “the tribal effect” – impassioned cultural and attitudinal characteristics that define a neighborhood and which either attract or deter people of like mind (which seems more prevalent now than ever before). Home security, child care/education, regional/hometown banking and environmentally-friendly products, as well as certain health/wellness services all come to mind as examples. Here, then, geodemographic systems can be a valuable component of the spatial analytics performed to factor those effects into targeting criteria.
In building predictive models, using geodemographic clusters as but one variable in the correlation algorithms that are built can be a valuable determinant in predicting likelihood to respond to an offer. Similarly, in evaluating the outcome of campaigns, having cluster definitions yields another analytical data point from which to draw attributing conclusions.
Seeing how your customer base is distributed geographically can also be valuable in understanding if and where there are target segments that are either over or under represented – which can impact a whole host of decisions, such as where to spend more strategically, or factor the nuances of the trade area into a more localized approach.
Surely, there are others. Key takeaway is that, just as individuals make decisions based on their own needs, belief systems, and circumstances, so do households in the way they function as an interrelated group of individuals both within the house and within a cluster of homes.
In the end, our target segment may still very well be individuals that we label something like Golden Year Guardians.Accuracy, actionability, and addressability are what count most – regardless of the data source.