Maximum difference scaling (MaxDiff)

Determining relative importance is a central task of market research. However, only in rare cases is it possible to ask direct questions about the essential features of products, markets or services:

The question “How important is … to you?” often results in an "inflation of aspirations", i.e. the respondents consider all attributes to be important. Moreover, systematic answer patterns employed in Likert scales or intercultural differences in how scales are used make it even more difficult to ask for preferences.

Maximum difference scaling (MaxDiff) is a ranking procedure that allows the best attribute out of several alternatives to be identified.

MaxDiff studies - when to be used?

MaxDiff studies can provide answers to all kinds of questions:

Which product features or services are most relevant to the purchase decision?
Which benefit statements/claims or names are most suitable?
Which image attributes are most suited to a brand?
Which packaging option is preferred?

A shower gel manufacturer, for instance, might be interested in which attributes of a shower gel are the most important to consumers buying such a product. MaxDiff scaling enables all possible attributes to be ranked using a simple questioning technique. For example, consumers could be asked whether it is the value for money, the natural ingredients, the cosmetic benefits or the fresh scent etc. that are most important to them.

How will MaxDiff studies be carried out?

The respondent is shown a series of screens with a selection of the attributes under investigation. On each screen, the respondent is required to choose the attribute they prefer the most and the one they prefer the least.

This results in a preference value being created for each attribute that reflects its relative importance in comparison to the other attributes.

MaxDiff surveys are carried out online or using the CAPI technique.

What are the advantages of MaxDiff studies?

This approach has a wide range of advantages:

a very simple technique for questioning respondents
better differentiation between the attributes compared to the level possible with rating scales (the respondent is "forced" to decide between the attributes)
no opportunity for scale-use bias (e.g. yea-sayer vs nay-sayer, extreme values vs average tendencies, intercultural differences etc.), meaning more valid results can be optained than with scales questions
can be carried out with a very large number of attributes
data is simple to interpret
delivers individual values for each respondent, allowing the results to be used for consumer segment analyses

MaxDiff and more

MaxDiff values are perfectly suitable for benefit-based segmentation – without the need to collect further data.

The results show how the market is segmented into consumer groups that give similar ratings to the alternatives tested. This allows strategically important market segments to be identified.

It is possible to generate statements that comply with the product or service requirements of different consumer segments. The findings allow recommendations to be made as to which segments should be addressed by the brand’s communications and can also be used to determine implications for the portfolio.