Key Drivers Analysis with MaxDiff

How do you measure the Key Drivers for your brand?

In order to effectively manage and position a brand in the market one of the important aspects is to understand what factors are important in influencing brand choice amongst consumers. The question that then arises amongst researchers is how best to measure these Key Market Drivers. This is quite a complex issue and there are a variety of ways to achieve it, with no single method being accepted as ideal. The answer to the question is therefore (the often annoying) "It depends".

At Leading Edge Research we have a toolbox of alternative approaches and will be able to advise you how best to research the issue.
The methods used can be divided into two basic groups; Explicit and Implicit techniques. In this article we briefly discuss the first of the explicit methods, MaxDiff or maximum difference scaling, which has become popular over the last number of years.

Key Drivers are generally based on Brand Attributes that get used to assess brand perceptions in the category. Attributes used can be classified in various ways and could include Performance or Functional attributes, Reputation or Image attributes, Price attributes, Personality attributes, Benefits attributes and Emotions.

The difference between Explicit and Implicit techniques.

Explicit

  • Direct questions - we ask the respondent directly what is important.
  • This could include a rating scale, a ranking of attributes or a trade-off approach
  • The attributes that emerge are those that are directly claimed to be important by respondents, therefore those attributes that will tend to operate at a very RATIONAL level.
  • These attributes provide consumers with a "reason to believe"

Implicit

  • A Derived analysis method that does not use direct questions. 
  • The importance of attributes is derived from brand attribute ratings or association.
  • Most analysis techniques rely on correlations between attributes and some overall measure of the brand - such as brand liking, equity or preference
  • These analyses will therefore tend to highlight SOFTER / EMOTIONAL or LESS RATIONAL attributes

Techniques include

  • Open ended / spontaneous questions
  • Rating scales - a weak measure not recommended
  • A ranking of attributes - easy and quite useful but has some drawbacks
  • Trade-off approach such as MaxDiff - the  best of the explicit approaches -  discussed below
  • Means End Chain method

 

Techniques include

  • Correlations
  • Penalties and Rewards analysis
  • Factor analysis and Regression analysis
  • Shapley Value analysis (LMG)
  • Random Forest analysis
  • Discrete Choice modelling also know as Conjoint analysis

The pros and cons of these techniques will be discussed in future posts

MaxDiff analysis is an explicit method that produces a very precise scaling of the relative importance of attributes. There tends to be high variation in the importance scores, compared to importance ratings for example, where many of the attributes tend to lump together with little difference in the importance scores.

The reason for the increased variability and improved scaling of scores is that respondents in MaxDiff are forced to trade-off or make choices between attributes. This reveals a scaling of importance scores that is more predictive of brand choice. The respondent task is as shown below:

MaxDiff_task

Respondents are shown 4 attributes at a time and asked each time which is the most important and which is the least important of the four. They are exposed to a number of such scenarios. The analysis allows us to determine the relative importance of attributes.


In a recent study we divided the attributes into 3 groups

  1. Product attributes or features
  2. Product Benefits
  3. Feelings or emotions

The basic outputs are average scores showing the relative importance of attributes

Graph

This analysis can be done in total or for selected target segments.

The importance scores are derived at respondent level. This allows us to further analyse the data. A very useful analysis is to use the scores themselves to derive a Benefit Segmentation. This allows us to identify segments of respondents / consumers who have the same patterns of importance scores and as such have similar requirements in terms of benefits expected from a brand in the category.

A Benefit Segmentation is one of the most actionable segmentations for marketers to implement and therefore a very useful way to segment the market.

Mail us to find out more

vincent@leadingedgeresearch.co.za