ALL DRIVERS OF SATISFACTION ARE NOT EQUAL: How do you find out what's really important?
by Paul M. Gurwitz, Ph.D., Managing Director, Renaissance Research & Consulting, Inc.
This article describes several methods for deriving feature importance and their advantages and disadvantages, including Stated Importance, Key Driver Analysis and The Kano Model. Renaissance Research & Consulting offers a model that has the combination of flexibility, validity, and ease of administration - RenSat+sm.
Every marketer wants to know what features of a product or service drive purchase. Over the years, several techniques have been used to get at that knowledge.
Originally, marketers simply gave respondents a list of features and asked them to rate their importance. The Stated Importance method was easy to administer and analyze, but it had two major problems:
- Because a respondent did not have to prioritize among features, it was hard to distinguish their importance – very often, all features turned out to be “very important”!
- Respondents’ answers were subject to “social desirability” or “yea-saying”: they rated the features the way they thought they “should”, which did not necessarily correspond to what actually drove product satisfaction.
Later, the technique of Derived Importance or Key Driver Analysis was developed. Respondents were not simply asked what was important; rather, they were asked to rate one or more brands on how well they performed on a series of product attributes. Statistical techniques were used to determine how strongly brand performance on each attribute predicted overall liking, and eventually usage, of the product; the indicators of that prediction represented the “true” importance of each feature.
Derived importance was a more valid indicator than stated importance; still, it had one major drawback. Conceptually, Key Driver Analysis was based on a simple, linear model of satisfaction: if a feature was important, then the more of it the product had, the better it did; the less it had, the worse, as in the following diagram:
In this conception, the only issue was the slope of the line: the steeper it was, the more important the feature. A “flat” line meant that the feature wasn’t important, since it made no difference in the level of satisfaction.
But what if consumers’ response to product features wasn’t as simple as that?
Prof. Noriyaki Kano developed an alternative to this simple satisfaction model. The Kano Model started with the original linear satisfaction model and expanded on it. It maintains that while consumers react to some features on a straight-line basis, others behave in a distinctly non-linear fashion:
- Basic Features are a cost of entry: if you don’t have them at least at a minimal level, then consumers won’t be satisfied. On the other hand, they won’t distinguish your product, either, so excessive attention to them probably won’t pay off: while no one will patronize a restaurant that isn’t clean, people won’t beat a path to a restaurant because it’s the cleanest in town, either!
- Exciters are the “extras” that set a brand apart. While consumers may not penalize a brand for not having them, these features boost satisfaction when a brand does have them, distinguishing them from competitors. These features should be cultivated and advertised, once a satisfactory level of the basic features is obtained.
The Kano Model broadens and deepens the original satisfaction model. However, as originally formulated, it poses two serious problems for researchers:
- It was designed as a stated importance model, in which consumers were asked to react directly to the importance of each attribute. As such, it was vulnerable to the problems of that approach: “yea-saying” and non-discrimination.
- The Kano model required that respondents be asked two questions about every feature:
- How would you feel about the product if it contained the feature?
- How would you feel about the product if it did not contain the feature?
This essentially doubled the length of the satisfaction battery, and along with it, the expense and the respondent burden.
In summary, the history of customer satisfaction modeling leaves marketers with a challenge: Could a customer satisfaction model have the flexibility and granularity of the Kano Model, with the validity and ease of administration of the Key Driver Model?
RenSat+sm: A Kano Approach to Derived Importance
We have been working for several years on incorporating the concepts of the Kano Model into the multivariate derived importance model. The result, RenSat+sm, is a customer satisfaction tool with the advantages of both models:
- The validity and discrimination of Derived Importance
- The “real-world” flexibility of the Kano Model
- Uses standard, single-question monadic rating data – no extra burden or expense.
How RenSat+sm Works
A RenSat+sm analysis is a four-step process:
- Monadic rating data are collected on attributes and features relevant to the product or service being tested, along with outcome data (e.g., overall rating, purchase likelihood, brand loyalty).
- The attribute ratings are factor analyzed. This step is necessary because even the most well-thought-out set of items will not necessarily capture the way the consumer views the category: the items as asked are only approximate measures of the terms the consumer actually thinks in. Factor Analysis identifies the way the consumer views the product or service, and how they relate to the questions you asked.
Factoring turns the original battery of attribute items into a smaller set of clearly distinct measures of the product’s performance, called factors. Unlike the original attribute items, the effects of each factor can be precisely measured statistically.
- The factors are first subjected to a traditional, regression-based derived importance analysis to ascertain the magnitude of the linear effects of each driver.
- Then, non-linear regression is used to identify each feature’s “Kano type”, and measure its impact on the chosen outcome. The non-linear terms of the regression are examined for statistical significance and direction. Factors with non-significant non-linear terms are traditional, linear key drivers; depending on the direction of their effects, those with significant non-linear effects are either “basic drivers” or “exciters”.
As in traditional derived importance analysis, RenSat+sm provides an overall importance weight for each factor; in addition, for “Basic Features” and “Exciters” it derives a second importance weight within its range of leverage. This “leverage weight” indicates how much an “Exciter” can boost product satisfaction if it’s present – or how much the absence of a “Basic Feature” can penalize a product.
A long-distance bus company wishes to know what drives rider satisfaction with the carrier. It asks a sample of travelers on its routes to rate its overall satisfaction with the carrier and its competitors, as well as on a list of 20 attributes that had been previously elicited in focus groups.
Factor analysis of the 20 attribute ratings reveal that they actually represented four distinct factors:
- LOW PRICE
- PLEASANT DRIVER
- ON-TIME PERFORMANCE
- ON-BOARD ENTERTAINMENT
A standard derived importance analysis suggests that, of the three significant factors, LOW PRICE was the most important:
However, further non-linear analysis shows a very different picture:
- LOW PRICE is a linear Key Driver: as the diagram below shows, the lower the price of the trip, the more satisfied passengers are with it.
- ON-TIME PERFORMANCE is a Basic Feature: its leverage works only on the downside. And its absence penalizes a carrier to a greater extent than low price helps it.
- ON-BOARD ENTERTAINMENT is an Exciter: its leverage only works on the upside; and its weight suggests that this “extra” could do more than anything else to distinguish a carrier from its competition.
The results of the analysis imply that the carrier should:
- First spend enough on maintenance and staffing to maintain parity with competition on on-time performance
- Provide free at-seat DVD players with a library of movies as a powerful “extra” to set themselves apart from their competition
- Stay competitive on fares, as long as it doesn’t negatively affect service.
By combining the flexibility and granularity of the Kano model with the efficiency and validity of the traditional derived importance model, RenSat+sm provides the marketer with a sophisticated new tool for fine-tuning the appeal of a product or service.
Other content shared by Renaissance Research & Consulting, Inc.
by Paul M. Gurwitz, Ph. D., Managing Director, Renaissance Research & Consulting, Inc.
Classical techniques of attitudinal segmentation, like Cluster Analysis, produce interesting groups that are not particularly useful. Targeting methods like CHAID produce groups related to an objective, and therefore useful, but they lack depth and are not interesting. This white paper presents a mix of the two methods - Hybrid Segmentation - as a method that produces segments that are both interesting and actionable. Read Article »
by Paul M. Gurwitz, Ph.D., Managing Director, Renaissance Research & Consulting, Inc.
One of the most popular ways of finding the correct price point is the Van Westendorp model. This article explains this method and proposes a new approach to price optimization developed by Renaissance Research & Consulting: VPlus+. Read Article »