Driver Analysis

Target-oriented evaluation and control of product developments and marketing activities

A driver analysis provides essential insights into the interdependencies between various characteristics of a product or partial performance of an offer and the overall assessment.

The results can be used to evaluate and control product developments and marketing activities in a targeted manner.

Driver analyses answer the following questions:

What significance does a partial performance – such as price or quality – of an offer have for the perceived overall evaluation?
What need for action or room for maneuver exists when individual product characteristics or performance dimensions are over- or under-fulfilled?
At what level of performance does the improvement of a driver, for example the product characteristic “sweetness” or the performance quality, not lead to any significant effect on the dependent variable, for example the willingness to buy the product or overall satisfaction with the offer?
At what level of a product characteristic or the performance level of an offer does the acceptance tilt, so that the overall satisfaction and performance are no longer evaluated positively?

Uncovering underlying drivers

The direct questioning of the relevance – often asked as “importance” or “ideal characteristics” of individual product features and services, leads to little reliable information. Even rather unconscious drivers are not recorded. In addition, direct queries often lead to an inflationary and exaggerated level of demands and generally provide little reliable and differentiating importance for the tested properties or partial performances.

For this reason, numerous statistical methods have been developed to determine the influence of various drivers through indirect queries and thus implicitly.

Experience has shown that Penalty-Reward Analysis (PRA) in particular delivers significantly better results here than direct and simple indirect methods and offers significant added value in the analysis of cause-effect relationships for product and offer development and optimization.

Linear interdependency

In driver analysis, Shapley value regression is most often used to determine linear relationships. Shapley value regression takes into account the multicollinearity (exists when two or more of the independent variables in a regression model correlate not only with the dependent variable, but also with each other) of drivers, leading to meaningful results.

Nonlinear interdependencies

Not all relevant drivers have a comparable, linear effect on the target variable, such as willingness to buy or satisfaction:

Some drivers, if not developed at all or too little, can only generate dissatisfaction, but not provide real increases in satisfaction. They are called basic factors.

Others have exactly the opposite effect: in the positive case, they increase satisfaction disproportionately, but in the negative case, they do not necessarily lead to dissatisfaction. They are called enthusiasm factors.

Drivers that work in both directions are called performance factors.

Kano-Model

This division into basic, performance and enthusiasm factors, which is based on the Kano model, is the subject of the penalty-reward analysis. It offers many advantages over linear driver analyses in terms of the quality and interpretability of the results:

The non-linear measurement or the subdivision of the drivers into basic, performance and enthusiasm requirements allows detailed statements for concrete and efficient improvements of a product or offer.
Based on the determined importance, the type of requirements and the achieved average performance assessment (performance), action-relevant partial characteristics or performance can be identified and compared with competitors.

The penalty-reward analysis is based on a Shapley value calculation to account for multicollinearity.

In addition to the quantified influence of the driver on the target variable (the driver strength), the “quality” of the driver is also identified. By taking into account the current performance per partial performance, it is possible to determine which drivers are particularly valuable for future measures.

How high is the part worth of a product feature? What influence does this feature have on the willingness to buy?

Relevant topics

SensUp® Packaging Test Product Testing