Over the last several years, we have seen something of a sea change in the business community’s increasing utilization of Max/Diff and related conjoint models. In their quest for better methods to assess complex sets of growth alternatives, professional researchers have turned to multi-attribute models.
The appeal of conjoint design approaches is the promise of greater discrimination between choices and “what if” scenario modeling of benefits or features. Early on, conjoint methods offered much more “realistic” choices for respondents, compared to traditional “laundry list” check offs of benefits, flavors, advertising claims, etc.
At the outset, the inherent assumption in the trade-off designs was that all the variables in a given choice alternative matter to all respondents. But, with experience came refinements, among them the recognition that there can be underlying segmentation of attribute valuation. The watershed was the development of adaptive conjoint (ACA), which takes into account variations in attribute or variable importance across respondents. Yet most choice-based designs still fundamentally assume that consumers place some value on every component included in the design.
In essence, all components collectively contribute to a choice. But the research history on choice-based models shows that significant proportions of respondents employ Lexicographic strategies when responding to a choice. Their choice is based primarily on a single dimension, such as price or brand or flavor. Depending on the study, 30%-70% of consumers have been found to be making choices on just a single attribute.
From this learning, there are several important insights that bear on understanding your organization’s brand landscape. First, the type and number of variables that drive choice can vary by category. Second, even within a single category, consumer choice strategies are not uniform. In many categories, segmentation of choice drivers is at play. As a result, some researchers routinely conduct Hierarchical-Bayes segmentation analysis of the choice based outcomes to be able to read them by how respondents made their choices (one attribute or specific combos). In a sense, this is meta data on respondent decision making – not just the attributes that are important, but how they are prioritized.
Which attributes drive choice in your brand’s landscape? Do the choice drivers vary by segment? Is that knowledge reflected in your choice-based research designs?
In an upcoming blog, we’ll discuss a related, crucial aspect of how consumers think about brand choices and how best to measure attribute imagery and delivery against attribute importance.