Data-Driven Insight: BCG’s MindDiscovery and More

Data-Driven Insight: BCG’s MindDiscovery and More

          
Article image

Data-Driven Insight: BCG’s MindDiscovery and More

  • Add To Interests
  • SAVE CONTENT
  • PRINT

  • Related Articles
    In This Article
    • Our consumer-research program incorporates MindDiscovery, a proprietary BCG approach featuring consumer workshops that make extensive use of psychologically derived techniques.

    • The initial qualitative exploration arms us with a comprehensive set of emotional, functional, and technical benefits in addition to market spaces.

    • We evaluate these using large-scale surveys that include at least three core sections: sizing, prioritization, and brand health.

     

    When a company embarks on a brand-centric transformation, our consumer-research program begins with a qualitative phase, seeking a comprehensive list of positioning options and market spaces that are owned, or could be owned, by a particular brand. Depending on category dynamics, this phase may incorporate in-context interviews, in-depth one-on-one interviews, or focus groups.

    It could also incorporate MindDiscovery, a proprietary approach developed by BCG's Center for Consumer Insight featuring consumer workshops that make extensive use of psychologically derived projective techniques. The diversity of stimuli and forms of expression helps alleviate many shortcomings of traditional focus groups. MindDiscovery avoids the limits of verbal expression, lowers social barriers, maintains a spectrum of views, and fosters creativity. It is particularly powerful for research on emotional aspirations. However, by combining creative stimuli with rational exercises, it can also build hypotheses about how functional and technical elements can deliver on an emotional promise.

    The initial qualitative exploration arms us with a comprehensive set of emotional, functional, and technical dimensions in addition to market spaces. We can then evaluate these using large-scale surveys (with sample sizes of at least 1,500). Respondent specifications and sampling methods vary by product category.

    Our surveys typically include at least three core sections: sizing, prioritization, and brand health.

    • Sizing. Questions that size potential emotional-positioning clusters and market spaces can be anchored on consumer segments, usage occasions, or any key dimension of value in the market (such as the percentage of meals consumed or the number of cars purchased in the past year). We then calibrate the results to external market data in order to evaluate the financial value of the spaces for the brand.

    • Prioritization. The heart of the survey, this lengthy section explores consumer expectations across the brand ladder. We find maximum difference scaling (MaxDiff) to be the optimal tool for prioritizing emotional and functional benefits. It allows for paired comparisons between all attributes tested and thus uncovers critical tradeoffs. For example, auto buyers may trade off “being worry free” against having the newest trendy car. An additional set of techniques (such as Kano modeling, traditional Likert scales, or concept testing) can complement the analysis of technical attributes.

    • Brand Health. This section of the survey is designed to compare the current state of a company’s brand to other brands in the product category by examining category-wide brand associations with each element of the brand ladder. We can see how well (or poorly) positioned a brand is for a particular emotional-positioning cluster or market space, and identify open-space opportunities or areas of vulnerability to competitive threats.

    Post-survey analysis leverages a series of statistical tools. For example, finding emotional-positioning clusters requires hierarchical Bayes analysis for MaxDiff scores, reduced by factor analysis. Brand association is explored through a modified chi-square test. Establishing links between emotional, functional, and technical statements may involve correlation analysis, t-tests, and regressions, or it may require more “black box” methods such as structural equation modeling and driver analysis.

  • Add To Interests
  • SAVE CONTENT
  • PRINT