Commonly measured marketing outcomes such as customer satisfaction, value for money, advocacy, and commitment are driven by the performance of a number of attributes or hypothesised drivers. In driving these outcomes invariably, some attributes are considered more important than others. Given limited resources, businesses would prefer to allocate resources to improving the performance on those attributes that are considered most important by the customer.

But how best do we ascertain this hierarchy of importance?

Essentially, there are two approaches; these are stated importance and derived or inferred importance.

The stated importance approach involves asking the respondent to rate the importance of a particular product or service attribute. The derived importance approach is based on determining the statistical association between the performance ratings for a product or service attribute and a broader performance criterion such as the overall ratings of the company.

The pros and cons of each of these approaches have long been discussed in the marketing annals and are summarised here:

 

Stated Importance

Advantages

• Very easy to ascertain since it simply involves the respondent providing a rating

• Researcher need not be familiar with multivariate techniques used in alternative approaches

• Simple interpretation for the client.

 

Disadvantages

• Insufficient variability amongst the attribute ratings which prevents the examination of how attributes are interrelated. For example, in response to being asked “how important is it to have knowledgeable staff? Interested staff? Committed staff?”, virtually all respondents reply “very important”

• Very costly and time consuming since every question needs to be asked twice – once for performance measurement and once for importance

• Lengthy questionnaires are likely to lead to problems with fatigue and completion rates.

 

But above and beyond these disadvantages with stated importance, all worthy of consideration, is the question of construct validity. That is, that the findings measure what they purport to measure and therefore can be generalised to the population. Too often, stated importance measures what is philosophically, politically or socially important to the respondent rather than what is important to bringing about outcomes such as satisfaction, value, advocacy or commitment. For example, a variable such as airline safety might be philosophically important to all airline travellers but irrelevant in choosing between Qantas and Virgin Blue for the Sydney/Melbourne route. The lack of construct validity is evidenced by the imperfect correlation found in studies between what is stated as important and what modelling derives as important.

Another example of this is where a customer rates the importance of ‘being compassionate’ as high because they feel they should. Yet when rating overall satisfaction, the impact of this attribute is not high in the consumers mind.

Conversely, it is also possible to have such a disconnect between stated and derived importance where the importance of the attribute is ‘understated’. An example of where this could occur is given by a customer rating the importance of ‘earning more than their peers’ as low because they think they should, yet this attribute has high impact on their overall satisfaction rating. Here, the derived importance of the attribute is greater than the stated importance.

Derived importance partially removes the risk that respondents will provide philosophically, politically or socially acceptable responses resulting in an inability to provide ratings that truly reflect the importance of an attribute in bringing about a marketing outcome. As a result, derived importance at least partly removes the risk that management will take action based on incorrect assertions.

 

Derived Importance

Advantages

Much fewer questions need to be asked without losing information. This saves time and money and avoids respondent fatigue.

Improved data quality is associated with the lower likelihood of respondent fatigue.

Through the use of statistical modelling, able to:

• determine which attributes contribute significantly to the overall construct, say, satisfaction and which don’t

• determine the hierarchy of importance of the significant attributes

• determine the amount of variation in say, overall satisfaction that is explained by these significant attributes

• statistically test the validity of the model.

 

Disadvantages

Requires knowledge of statistical modelling

 

What Should We Be Earnest About?

For management, surely the relative importance of an attribute is a more crucial piece of information than the relative performance. If an attribute is of little importance then its relative performance will not be what management needs to focus on. Therefore, how marketing research identifies the hierarchy of importance is vital. With the proliferation of multivariate analysis there seems little justification for asking respondents to score the importance of an attribute. Today, marketing research vendors should need to justify reporting stated importance.

Of course, the author acknowledges that she is only stating how important importance is!