Identification of Critical Attributes of Satisfaction in Products and Services: an Alternative to Importance - Performance Analysis
1 INTRODUCTION
As the competition for new markets and customers grows among businesses, customer satisfaction becomes a key factor for long-term business success. Satisfied customers are loyal customers and ensure a lasting cash flow in the future. According to Reicheld and Sasser (1990), an increase in customer loyalty by 5% can increase the profit of a business by 100% due to the fact that satisfied customers purchase the products of a company more often and in greater quantities. Anderson and Mittal (2000), studying companies that are part of the Swedish Customer Satisfaction Barometer, found that an increase of 1% in the customer satisfaction index was associated with 2.37% increase in the return over investment, while a decrease of 1% was associated with a decrease of 5.08% in the return over investment. These results show that while increasing customer satisfaction is important, avoiding customer dissatisfaction is critical. But how can a company continuously satisfy its customers? Satisfaction is related to the fulfillment of implicit and explicit customer needs by the totality of the product attributes. So, it becomes important to find out how the performance of different attributes impact on customer satisfaction.
In a competitive environment it is not enough to find out the importance of the different product attributes. It is also important to follow changes in customer needs and constantly evaluate the product’s competitive position. The impact of the different attributes on customer satisfaction changes over time either due to the fact that customers get used to them, due to the offer of substitute products or due to movements of competitors introducing improvements in the existing products (Tontini, 1996).
Most of the traditional techniques aimed to find out the relative importance between attributes assume that customers have previous knowledge about the product and its attributes (Deszca; Munro and Noori, 1999), hindering the introduction of innovations. Besides, they assume that there is a linear relationship between attribute performance and customer satisfaction, what may lead to wrong decisions about which attributes should be improved or offered to increase customer satisfaction (Huiskonen and Pirttilä, 1998). Kano Model of Customer Satisfaction (Kano, Seraku, Takahashi, & Tsuji, 1984) identifies this non-linear relationship by classifying product and service attributes in basic, performance and excitement factors, but it does not take in consideration the present level of performance of the attributes, being limited as a tool to find out which attributes should be improved.
This work has as objective to evaluate the limitations of the Importance - Performance Analysis (IPA) and of the traditional Kano Method, proposing a new method to identify the impact of variation in attributes’ performance on customer satisfaction. The paper is structured as follows: in section two we explain the methodology used to illustrate the proposed method and to analyze the limitations of IPA and Kano Method. In section three we explain the IPA and explore its limitations. In section four we explain the traditional Kano Method. In section five we explore the possible mistakes that a company may incur by not considering the asymmetrical relationship between attribute performance and customer satisfaction. In section six we introduce and illustrate the new method. Section seven presents the conclusion of the work.
2 RESEARCH METHODOLOGY
A survey with customers of pizzerias was used to illustrate either the limitations of the Importance - Performance Analysis and of the traditional Kano Method, as well as to illustrate the proposed alternative method to identify the critical attributes of success and improvement opportunities. Although any product or service could be used in the study, the service “pizzeria” was chosen because it is of general knowledge to all respondents. The survey was conducted with a random sample of 161 students of the Regional University of Blumenau (FURB). Since the students live in different areas and frequent different and competing pizzerias, the sample was considered adequate for the purpose of this study. In order to increase the reliability of the survey, all respondents accepted to participate in the research voluntarily, taking enough time to answer the questions with the best of their judgment. Four attributes were studied: cleanness, courtesy, choice of pasta besides pizza and diversified filled border, i.e., filling the border with the same topping of the pizza. “Diversified filled border” is a new attribute, not offered in the market. The attributes were identified by conducting a focus group and were specifically chosen to include the different categories of the Kano Model. From now on we call the choice of pasta as “pasta offering” and the diversified filled border simply as “filled border”.
Four parts composed the survey questionnaire. In the first part, the respondents were asked to rate their satisfaction with imaginary situations of sufficiency and insufficiency performance for each attribute in a modified Kano questionnaire (discussed in section six). Then, using a Likert scale ranging from very dissatisfied (-3) to very satisfied (+3), they were asked to rate their satisfaction with the performance of the attributes in the last visited pizzeria. Also, the respondents where asked to rate the importance of each attribute in a scale ranging from 1 to 5, being 1 “not important at all” and 5 “very important”. At last, it was asked the frequency of eating in pizzerias and the name of the last visited pizzeria.
Figure 1 illustrates the frequency distribution for the last pizzerias visited by the respondents of the survey.
Figure 1 – Frequency distribution for visited pizzerias
From now on we will call “Pizzeria A” the reference pizzeria and “Pizzeria B” its largest competitor.
3 THE IMPORTANCE PERFORMANCE ANALYSIS
The Importance - Performance Analysis, introduced originally by Martilla and James (1977), allows a company to have a vision about which attributes of its product or service should be improved to become more competitive in the market. Typically, data coming from customer satisfaction surveys are used to build a bi-dimensional matrix, where the importance is shown by the y-axis and the performance of the attribute by the x-axis (figure 2).
Figure 2 – Traditional Importance – Performance Analysis
The matrix is divided in four quadrants. An attribute located in Quadrant I have high importance and high performance, representing a possible competitive advantage. In this case the company should "maintain the good work". An attribute that has high importance, but low performance, should receive immediate attention (Quadrant II). The Quadrant III contains the attributes with low importance and low performance, not being necessary to concentrate additional effort on them. The Quadrant IV concentrates the attributes with high performance but with low importance. In this case the company can be wasting resources that could be better used in another place.
3.1 EVALUATING ATTRIBUTES’ PERFORMANCE
In this paper, the performance is evaluated by the average satisfaction with each attribute. Table 1 displays the average satisfaction and the p-values for Pizzeria A, Pizzeria B, market average and the relative position between pizzerias A and B for each studied attribute.
Attribute / Pizzeria A / Pizzeria B / Market average / A/B / p-value (A=B) / p-value(A = Market)
Cleanness / 1,45 / 1,40 / 1,51 / 104% / 0,400 / 0,22
Courtesy / 1,18 / 1,69 / 1,36 / 70% / 0,055 / 0,25
Pasta Offering / 1,09 / 1,51 / 1,00 / 72% / 0,091 / 0,38
Filled Border / -0,51 / -0,53 / -0,51 / 96% / 0,480 / 0,50
Table 1- Average satisfaction for the studied attributes
Table 1 shows that customers of Pizzeria A have the same satisfaction than the market average in all attributes (p > 0,1). Comparing Pizzeria A with Pizzeria B, its largest competitor, we see that customers of Pizzeria A are less satisfied than customers of Pizzeria B in the attributes “courtesy” and “pasta offering” with at least 90% of confidence. Although there is not enough data to indicate accurately the absolute level of satisfaction with the attributes, the results show that the average satisfaction with “filled border” in the market is slightly negative. Since the “filled border” is not offered by any of the competitors at the moment, this dissatisfaction may indicate a possible unarticulated need that could become a differential for the first competitor who offers it.
If an isolated analysis of the satisfaction results is used as an indicator of the areas or attributes that should be improved, table 1 suggests that Pizzeria A should improve its “courtesy” and its “pasta offering”.
3.2 EVALUATING ATTRIBUTES’ IMPORTANCE
Two ways are commonly used to estimate importance of the attributes: stated importance and statistically inferred importance.
In the stated importance method customers are asked to rate the importance of the attribute, typically ranging from “not important at all” to “very important” in a Likert scale. This method has some limitations:
a) Consumers tend to give higher importance to attributes that represent the basic functions of a product or service. For instance, "keeping the temperature low" is the basic function of a refrigerator. If asked to rate the importance, consumers may give high importance for this attribute, however, as all competitors reach a satisfactory level, the purchase decision will be related to other attributes as design, color, size, etc. (Garver, 2003, p.458).
b) Usually, stated importance tends to have low discrimination power and the customer tends to find everything important. Another way of estimating importance is to ask customers to rank the attributes in order of importance. This method increases the discrimination power but it may be difficult for a respondent to rank 15 or more attributes due to the timing required and the number of ties in importance.
The stated importance for the attributes studied in this paper is shown in table 2. “Cleanness” was considered the most important attribute, followed by “courtesy”, “pasta offering” and “filled border” respectively. But, to what extent could “cleanness” be considered a differential for attracting customers? Would not be “courtesy” more effective in satisfying them? Besides, since the value “3” corresponds to “important” in the scale, the tendency of customers to find everything important is evidenced.
Attribute / Average ImportanceCleanness / 4,77
Courtesy / 4,29
Offering Pasta / 3,02
Filled Border / 2,94
Table 2 – Stated importance
For the statistically inferred importance, customers are asked to rate both their satisfaction with the current performance of the different attributes and their general satisfaction with the product or service under study. Then, a linear multiple regression equation is adjusted between the satisfaction with the individual attributes (independent variables) and the general satisfaction (dependent variable). Equation 1 shows the regression equation for the case studied in this paper.
General Satisfaction = b0 + b1 Cleanness + b2 Courtesy + b3 Pasta + b4 Filled Border (Eq. 1)
where Cleanness, Courtesy, Pasta and Filled Border are the satisfaction level for each attribute.
The higher the values of b the higher the impact of increasing satisfaction with one attribute in the general satisfaction, and the higher the importance of the attribute. This method eliminates the tendency of customers to find all attributes important and discriminates better the relative importance between them. However, this method is not free from deficiencies. The linear regression assumes that: a) the data have normal distribution, b) the relationship among the variables is linear and c) the multicollinearity among the independent variables is low. According to Garver (2002), these premises are frequently violated in consumer's satisfaction data. Customers tend to say that they are "satisfied" when actually they are in a neutral state, leading to a bias in the data. Also, the relationship between the satisfaction with an attribute and general satisfaction is not linear (Ting & Sheng, 2002)
Table 3 displays the result of regressing the satisfaction with the attributes against the general satisfaction with the pizzerias. The most important attribute was “pasta offering”, followed by “courtesy”, “cleanness” and “filled border”. These results are different from those of stated importance (table 2). This difference is due to the fact that stated importance for each attribute is strongly dependent on customers’ perceptions arising from personal and social aspects. For instance, in most cases “cleanness” will tend to be considered as the attribute with highest importance due to the negative image consumers have in case of its absence.
Attribute / b Coefic. / p-value / Significanceb0 (constant) / 0,87 / p < 0,001
Cleanness / 0,16 / p = 0,037
Courtesy / 0,27 / p < 0,001
Pasta Offering / 0,30 / p < 0,001
Filled Border / 0,02 / p = 0,69 / Not significant
R2 / 0,46
Table 3 – Statistically inferred importance
On the other hand, statistically inferred importance depends on the previous experience of the customers with the product or service. In the pizzerias case, as all competitors reached a satisfactory level in “cleanness”, the variation in satisfaction is small, making the coefficient b1 smaller and this attribute less important. Also, since the respondents of the survey do not know “filled border”, they are neither satisfied nor dissatisfied, making the coefficient b4 insignificant and this attribute not important at all.
The divergent results of the stated and the statistically inferred importance methods put in check the decisions to be made during the design of a product or service. Garver (2003); Matzler and Sauerwein (2002) relate a way to classify the attributes through an analysis using the results of the two methods. Attributes that receive high importance in both methods are considered key attributes. Similarly, attributes that receive low importance in both methods are considered secondary. Those attributes that receive high importance in the stated importance method and low importance in the statistical method are considered basic. Those that receive low importance in the stated importance method and high importance in the statistical method are considered as "amplifiers."
Using the combined method for the pizzerias case, we discovered that “cleanness” is considered basic, “courtesy” is a key attribute, “pasta offering” is an amplifier attribute and “filled border” is a secondary one (figure 3). In this case, the average of the b coefficients (0,19) was used as the dividing line between high and low importance in the statistical method, and "important" (3) as the dividing line in the stated importance method.