Naked Attitudes
Speak to most DM and CRM managers about using attitudinal research in customer management and you’ll probably get one of two responses: either “It can’t be done” or “We’ve segmented the market by attitudinal group and we’re now working with our research agency to replicate this segmentation on our marketing database”.
The first view is clearly myopic (sic) whilst the second is a simply a trap everyone is falling into simply because everyone else does it that way. Well, almost everyone.
Regretfully, neither of these approaches actually adds much benefit to one-to-one marketing effectiveness. But try to point this out and it is as if you are saying the emperor has no clothes. “That can’t be true, can it?”
The problem the practitioners have stems from trying to project attitudinal segments onto a marketing database. To do this requires two criteria to be fulfilled. Firstly, you need to able to find a set of variables on the database that are statistically significantly correlated with membership of each specific attitudinal segment. Secondly, you need to achieve this with a limited sample size because the cost of surveying everyone is usually prohibitive (unless your database is small to start with or is purely on-line).
So the fact that many have tried and failed to achieve a satisfactory result is hardly surprising. It is like to trying to refine crude oil with a cheese grater. As soon as someone speaks of “replicating attitudinal segments on the database” much of the richness that exists in the attitudinal research is instantly discarded in favour of (typically) six or eight segment flags and the geodemographic (and, occasionally, psychographic) variables that describe the people within them. This makes no sense.
What companies should be doing is looking for the key attitudinal drivers that motivate people to behave as they do. These drivers determine which segment someone is in but instead of flagging someone with a segment label they can be scored on their attitudinal drivers. This creates extra variables for the marketing database which are much more actionable and much more powerful than simple segment flags. Those might be added as well but mostly for internal reporting purposes but for targeting they are, and always will be, extremely blunt tools.
So what are these drivers? Take a simple example. In the gardening category, we might well find attitudinal segments such as “Horticulturalists”, “Ladies who Lunch”, “Aspiring Mothers”, “Grass Cutters” and “Dreaders”. The extent of their engagement in the category and their attitudes and reasoning for having that engagement is reflected in the segment titles. But how confident can we be that someone is in a particular segment? At best it is with only a partial confidence that is often less than 50%.
However, the drivers (or factors) that determine someone’s likelihood to be in a particular segment can be derived through a factor analysis of the market research data. In the gardening sector we know from working with our clients that there are three key drivers: Cost; Recreation; and Esteem. Usage and Attitude research enables us to identify these drivers and, more significantly, how important each one is to each individual respondent in determining the attitudinal segment to which they are likely to belong. For example, the only reason that “Dreaders” may engage in the sector is for reasons of cost. They derive no pleasure in gardening in itself nor wish to garner any outward appreciation from others from the results of their endeavours.
“Horticulturalists”, by contrast, score highly on all counts. They take pride in growing plants from seeds because of the money they save, enjoy the recreation for personal pleasure and take part in flower shows and other events to gain external admiration.
Therefore, by knowing these three simple factors we can build many hypotheses about the types of customer measures we can use to score the importance of each factor to each individual on the database. We can then use those scores as additional targeting variables. A “Horticulturalist” will often be as interested in saving money on a lawnmower as a “Dreader” but the “Ladies who Lunch” will be just as much (and probably more) interested in making sure it’s easy to operate as they are in the cost.
By having each person scored on the likely importance to them of each of the Cost, Recreation and Esteem drivers means we can use those scores to develop targeting algorithms (and analyse campaign responses) using variables that reflect a customer’s attitudes but do so person-by-person, not just bluntly segment-by-segment.
This makes much more sense because, as all surveys and all databases contain imprecise information, forcing people into specific attitudinal groups will, by definition, put many people in the wrong boxes. This mutes the impact of using attitudinal data to increase response rates. Using the drivers avoids such artificial segmentation and replaces it with simple probability score variables. Look for these instead and response rates will grow by at least 50% every time you do so.