The Importance of Reliability and Customization from Goods to Services

January 2003
Volume 10 • Number 1


The Importance of Reliability and Customization from Goods to Services

by Michael D. Johnson, University of Michigan Business School, and Lars Nilsson, Karlstad University

Although there is a substantial body of research on quality, disagreement remains as to the effect of reliability, or things gone wrong, as opposed to customization, or things gone right, on customer satisfaction with goods and services. Service quality researchers argue that reliability is relatively more important for services because of the nature of service production compared to goods production. In contrast, customer satisfaction researchers argue that a service firm’s ability to customize its service to individuals makes customization relatively more important for services than for goods. The goal of this article is to provide insight into this debate through an analysis of firms and industries measured in the American Customer Satisfaction Index database. The authors’ results provide broad-based support for the argument that reliability is relatively more important as the service component of an offering increases, while customization is relatively more important for manufactured goods. Implications for quality practice are discussed.

Key words: customization, customer satisfaction, goods, quality improvement, reliability, services


Central to a firm’s customer orientation is the improvement of those aspects of quality that are most important to customers. Quality experts distinguish between two general types of quality: 1) customization, or the degree to which a good or service provides key customer requirements, and 2) reliability, or how reliably these requirements are delivered (Deming 1981; Juran and Gryna 1988). The methods and processes used to improve these two quality types can be quite different. When improving customization or “things gone right,” greater emphasis is placed on customer research as a basis for understanding customer needs. When improving reliability or “things gone wrong,” greater emphasis is placed on operations and the need to failsafe existing products and processes. It is important, therefore, to understand the relative importance of customization vs. reliability when allocating resources for quality improvement.

The issue is particularly relevant as the service component of an offering increases relative to the physical goods component. There remains a general disagreement as to which quality dimension becomes more important to maintain or improve as one moves from goods to services. Scholars in the services management tradition argue that the coproduction process that typifies services makes reliability a particularly important quality dimension (Grönroos 1990; Zeithaml, Parasuraman, and Berry 1996). Unlike pure goods, pure services are coproduced with customers at a time and in a place of the customer’s choosing. And because service production involves more of the human resources of the firm and customers themselves, it adds greater inherent variability to the service production process. Thus reliability should be very important to maintain or improve. Others argue that the same coproduction process makes customization relatively more important in determining customer satisfaction for services (Anderson, Fornell, and Rust 1997; Fornell et al. 1996). Because many services are personnel intensive and customized to suit heterogeneous needs, customization is more important for services than for manufactured goods.

However, there is no broad-based evidence of the relative importance of customization vs. reliability across the goods-to-services continuum. The authors’ goal is to provide a direct test of the relative impact of the different quality dimensions on customer satisfaction. They use American Customer Satisfaction Index (ACSI) data from as many as 188 firms and 30 industries over the period 1994 through 1998 to test their hypotheses. They first contrast goods vs. services, and then examine firms and industries across four categories along a goods-to-services continuum, to test their predictions. The next section of the article develops the arguments and describes the available evidence on each side of the debate. The authors then describe the ACSI data and analyses used to test the hypotheses, report the findings, and discuss their implications for quality practice.


Over the course of history, the definition of quality has evolved and changed. It has been equated with excellence, value, conformance to specifications, and benchmark superiority. New definitions have not replaced old definitions; rather, all the quality definitions continue to be used today (Reeves and Bednar 1994). More common in the latter part of the 20th century is the definition that quality is conformance to requirements and customer specifications (Crosby 1980). One of the most well-known concepts of quality is J. M. Juran’s concept of fitness for use (Juran and Gryna 1988). Reeves and Bednar (1994) view Juran’s definition of quality as a combination of quality as excellence and quality as conformance. Juran emphasizes that quality is the extent to which a product successfully serves the purpose of the user. Customers’ view of quality is similarly derived from two distinctly different dimensions: product performance and freedom from deficiencies. Product performance is the degree to which the product’s specifications are customized to meet the needs of any given customer. Freedom from deficiencies is simply how reliably the product meets its specifications. Ishikawa and Lu (1985) make a similar argument when they separate quality into “backward-looking” and “forward-looking” components. Defects and flaws in quality are called backward-looking quality, while forward-looking quality is that which becomes a product’s sales point. For the purpose of this discussion, the authors refer to these two components of quality as customization and reliability.

Although these features may compete with each other in the marketplace, as when one competitor excels on reliability while another excels on customization, they are only conceptually independent. Reliability is, at some level, a necessary condition for differentiation and customization to exist (Fornell and Johnson 1993). Empirically, the two dimensions are closely related components of overall quality. The advantage of this two-dimensional view of quality is that it allows one to make comparisons across very different firms and industries. It is arguably more appropriate to use more abstract, inclusive dimensions when making cross-industry comparisons between relatively “noncomparable” goods and services (Johnson and Fornell 1991). Within any given industry, firm, or even market segment, the quality dimensions that drive customer satisfaction and loyalty are many and varied (Feigenbaum 1991; Garvin 1984; Gustafsson and Johnson 1997; Parasuraman, Zeithaml, and Berry 1985). Yet all the dimensions may be viewed, at some level, as falling under the categories of customization and reliability.

Although the definitions of quality described thus far and the distinction between reliability and customization apply to both goods and services, they have evolved primarily from the study of manufactured goods. Services management has evolved into a research field of its own that provides unique insights into the measurement and management of service quality. Services have several unique qualities relative to physical goods. Services are more intangible than goods, making them hard and sometimes impossible to count, measure, inventory, and test (Grönroos 1990; Parasuraman, Zeithaml, and Berry 1985). This often makes it difficult for customers to understand service quality and, as a result, more difficult for firms to understand how consumers perceive and evaluate a service (Zeithaml, Parasuraman and Berry 1990). Unlike goods, where production and consumption are typically separated by time and space, services are coproduced at a time and place of the customer’s choosing (Parasuraman, Zeithaml, and Berry 1985). The inseparability of production and consumption for services means that service reliability is more outside the control of the firm. Finally, service production differs from goods production in that the coproduction process involves more of the human resources of the firm and customers themselves (Grönroos 1990). There is simply a higher ratio of people to inanimate objects in the “service factory.” As a result, services exhibit higher variances that cannot be controlled by the service process.

The Reliability Argument and Evidence

Service quality and customer satisfaction researchers have raised two very different arguments regarding the implications of service production vis-à-vis goods production for customization and reliability. As argued earlier, the simultaneity of production and consumption combined with the greater ratio of human (both employee and customer) involvement creates more inherent reliability problems for services vis-à-vis goods. This suggests that improving reliability or minimizing things gone wrong, as opposed to improving customization, is relatively more important for services. What customers expect service companies to do is provide the fundamentals, or a service free from deficiencies (Parasuraman, Berry, and Zeithaml 1991a). Both the service quality literature and the customer satisfaction literature provide some support for this argument.

Primary support for the importance of service reliability in the services quality literature comes from studies using the SERVQUAL survey methodology (Parasuraman, Zeithaml, and Berry 1985; 1988). The SERVQUAL method measures five dimensions of service quality:

  1. Tangibles: The appearance of physical facilities and equipment
  2. Reliability: The ability to perform the promised service dependably and accurately
  3. Responsiveness: The willingness to help and provide prompt service
  4. Assurance: Employee knowledge and courtesy, and ability to inspire confidence
  5. Empathy: Caring, individualized attention

The attributes of service reliability in the SERVQUAL survey include:

  1. The degree of fulfilled promises
  2. The degree of interest in solving one’s problems
  3. Whether the services are provided right the first time
  4. Whether the services are provided at the time they are promised
  5. The existence of error-free records

The traditional SERVQUAL approach asks customers to rate the level of performance or excellence they would expect to see on each attribute of each service quality dimension and then rate the level of performance they actually receive. Those dimensions with the largest gap between expectations and performance are the most important to improve.

In their review of the SERVQUAL research, Parasuraman, Berry, and Zeithaml (1991b) conclude that reliability is consistently the most important dimension, or largest “gap,” to improve across service industries. Responsiveness is consistently second in importance, while tangibles are consistently the least important dimension to improve. Researchers have criticized the SERVQUAL approach on methodological grounds, including the validity of the five dimensions, the use of direct expectation or importance measures, and the use of difference scores (see Bateson and Hoffman 1999). Berry, Parasuraman, and Zeithaml (1994) subsequently demonstrated the generalizability of their findings using a different methodology (allocation of 100 points among the five dimensions) to identify the most important dimensions to improve. Their results again support reliability as the most important service quality dimension (32 percent) followed by responsiveness (22 percent), assurance (19 percent), empathy (16 percent), and tangibles (11 percent).

These results, albeit important, do not directly address the question that motivates this research. The authors expect that reliability is relatively more important to customers than is customization as one moves from physical goods to services. The SERVQUAL studies focus on services; they do not include benchmarks for physical products. Moreover, reliability is but one of five SERVQUAL dimensions; reliability is never compared to customization per se. Another potential limitation of the SERVQUAL research is that it involves one form or another of direct importance measures (direct scale ratings or point allocation methods). Gustafsson and Johnson (1997) argue that statistically derived importance measures provide more objective estimates of the impact that a given quality dimension (such as customization or reliability) has on overall satisfaction.

Research using the national customer satisfaction barometers or indices provides more direct comparisons of goods and services. The Swedish Customer Satisfaction Barometer (SCSB) (Fornell 1992) provides data on satisfaction across approximately 30 industries and 130 firms. The ACSI (Fornell et al. 1996) provides data on approximately 35 industries and 200 firms. These surveys find satisfaction with services falling consistently and significantly below the level of satisfaction with more physical goods. The authors contend that the satisfaction gap is most likely a reflection of the inherent reliability problems that plague service production. If customization were a service provider’s relative strength, service satisfaction should be higher than goods satisfaction, not lower. If reliability were a service provider’s relative weakness, it would suggest lower satisfaction for services. However, the totality of the evidence is indirect.

The Customization Argument and Evidence

That reliability becomes more important than customization as one moves from goods to services has been challenged of late. Huff, Fornell, and Anderson (1996) argue that reliability is likely to drive overall quality when there is meaningful variation in defects between competing products and customers are able to differentiate the variation. The intangible, subjective nature of service performance makes the customers’ ability to differentiate service variation more difficult. Huff, Fornell, and Anderson (1996) thus conclude that reliability should be relatively more important for customer satisfaction with goods.

Anderson, Fornell, and Rust (1997) further argue that service production offers important advantages over goods production. The coproduced nature of a service allows for intensely personal and customized services that suit a very heterogeneous set of needs (see also Colgate and Danaher 2000; Grönroos 1990; Bateson and Hoffman 1999). Effective service firms find ways to take advantage of the inherently flexible nature of service production to more than compensate for the problems of delivering consistent and predictable levels of service quality. This suggests that customization is relatively more important than reliability for services when compared to goods. Accordingly, Hoffman and Bateson (1997) argue:

“Producers of goods typically manufacture the good in an environment that is isolated from the customer. As such, mass-produced goods do not meet individual customer needs. Since both the customer and the service provider are involved in the service delivery process, however, it is easier to customize the service based on the customer’s specific instructions” (Hoffman and Bateson 1997, 34-35).

The emergence of service recovery systems (Smith, Bolton, and Wagner 1999) is consistent with this argument. Service recovery might reduce the negative impact of reliability, while increasing the positive impact of customization, on service satisfaction. A study by Bolton (1999), however, concludes that, for a majority of customers in both a restaurant and hotel setting, customer satisfaction and repurchase intentions decrease after a service failure and recovery encounter.

The primary evidence in support of the customization argument comes from Fornell et al.’s (1996) study of the 1994 baseline ACSI data. The ACSI model measures quality using an index of three survey questions (overall quality, customization, and reliability). The Quality Report is a driver of customer satisfaction in the model. The measurement loadings (the correlations between the measures and the overall Quality Report) suggest that customization is more important among the service industries studied than among the manufactured goods industries. The average loadings for customization were 0.909 for services compared to 0.898 for manufactured goods, leading the authors to argue that customization is more important for services.

Yet there are several problems with this evidence. The difference between the loadings is quite small and the significance not reported. It is also problematic to view the customization loadings in isolation. Any conclusions regarding customization must take into account the loadings for reliability, which were also not reported. Finally, standardized loadings are not the same as effect sizes (impact scores). More direct tests of the arguments surrounding customization and reliability require an analysis of impact or effect size rather than correlation. In the end, while there are good arguments on each side, there is simply no broad-based evidence as to whether customization or reliability becomes more important when moving from tangible, physical goods to intangible, coproduced services.


Market offerings are best described as falling along a goods-to-services continuum. After describing the goods-to-services continuum in more detail, the authors posit the research hypotheses to be tested in their empirical study.

The Goods-to-Services Continuum

Most of the products in today’s economy consist of some “good” as well as “service” components (Bateson and Hoffman 1999; Mittal, Kumar, and Tsiros 1999). Even in a traditional service/retail organization like McDonald’s customers receive goods (such as a hamburger, fries, and a soda) and services (such as a short waiting time and friendly service). Although the ratio of goods to services in the offering is difficult to quantify, there are categorical distinctions in the literature that one can use (Martin and Horne 1992; see also Kotler 2000). There are four categories that work well for the ACSI industries in the authors’ subsequent study: 1) pure goods (food products, soft drinks), 2) core goods with accompanying services (cars, computers), 3) core services with accompanying goods (airlines, hotels), and 4) pure services (phone service, banking). In this analysis of the ACSI data, the authors first examine differences between both firms and industries with respect to their primary classification as good (categories 1 and 2) or service/retailer (categories 3 and 4) based on SIC code classifications. This is consistent with earlier ACSI research (Fornell et al. 1996). They then examine firm-level differences in more detail using the four level goods-to-services continuum.


The authors’ discussion leads to a series of research hypotheses that are tested in their empirical study. The first two hypotheses posit main effects for both reliability and customization on customer satisfaction.

Hypothesis 1: Perceived reliability has a positive impact on customer satisfaction.

Hypothesis 2: Perceived customization has a positive impact on customer satisfaction.

While these hypotheses are already well supported (Fornell et al. 1996), it is important to test them again in the context of the potentially moderating effects of the distinction between goods and services.

The third and focal hypothesis is motivated by the argument that the coproduction of services creates greater inherent reliability problems for services as compared to goods. The authors predict that improving reliability has a relatively greater impact than customization when moving from goods to services.

Hypothesis 3: The relative impact of reliability vs. customization is greater for services as opposed to goods.

The alternative hypothesis is that, because customization is easier for services and service reliability may be more difficult for customers to detect, the opposite holds. That is, the impact of reliability vs. customization decreases from goods to services. The authors’ focus is on relative changes in the impact of reliability and customization because service researches agree that quality may be more difficult to judge for services (Parasuraman, Zeithaml, and Berry 1985). This is due to the intangibility of services; there are fewer tangible cues available to judge quality. Thus the impact of both customization and reliability may decrease somewhat, in an absolute sense, from goods to services.


The hypotheses are tested using data from the ACSI survey (Fornell et al. 1996). The ACSI model is estimated for each of approximately 200 firms annually and based on a random sample telephone survey of approximately 250 of a firm’s customers. The authors’ interest is in the overall satisfaction index and the survey ratings for customization and reliability. Specifically, they examine the impact that customization and reliability ratings have on the satisfaction index to determine the relative importance of each quality dimension. The satisfaction index is a weighted average of three survey ratings: 1) an overall rating of satisfaction; 2) the degree to which performance falls short of or exceeds expectations; and 3) a rating of overall performance relative to the customer’s ideal good or service in the category. Each measure is an overall evaluation of the good or service. The 1- to 10-point scale ratings for these measures are combined into a weighted average and rescaled to provide a 0- to 100-point satisfaction index (see Fornell et al. 1996 for details). Our measures for customization and reliability are on their original 1- to 10-point scales (where 1 = poor customization or reliability, and 10 = excellent customization or reliability).

Sample of Firms and Industries

The authors use a sample of both firm- and industry-level data from the 1994 to 1998 ACSI survey to test the hypotheses. Not included are public or government agencies because of the relative lack of consumer choice. Outliers were deleted from the samples by building separate regression models for reliability and customization on satisfaction using the industry-level data as observations and studying the studentized residuals. If an industry in a given year exceeded the specified limit (greater than +/- 2.0; see Hair et al. 1995), it was deleted. For consistency, the firms within these industry-level observations were similarly deleted from the firm-level observations. The deletions included personal computers, household appliances, broadcasting, and long-distance phone service for 1997 and 1998, the U.S. Postal Service (parcel delivery) for 1997, and utilities for 1998.

The primary focus of this analysis is the firm-level sample. A small number of companies were dropped from the sample because they were either dropped from the ACSI in the midst of the time frame or the structure (composition) of the industry changed. Altogether 97 such observations were deleted representing 43 different companies. The final sample contains a total of 823 firm-level observations, which includes 141 to 188 firms in any given year. The five years of data were stacked for purposes of the empirical analysis and a five-level categorical variable was included to control for differences from year to year. In a small number of cases values for a variable were missing. In these cases (six cases out of 823 observations), the authors replaced the value with the previous year’s value for the firm on the variable. This rule is consistent with the observed stability of the ACSI measures over time.

The final industry-level sample includes 25 to 30 industries in any given year. There are a total of 140 industry observations over the five years. Table 1 shows the 30 industries classified along the four-level goods-to-services continuum. The two-level classification (goods vs. services) is based on SIC code classifications previously used in ACSI research (Fornell et al. 1996). Subdivision of the goods and services into the four-level classification (pure good, core good with services, core service with goods, core service) was based on discussion and agreement between the authors. The only two industries for which the classification was not so clear were household appliances and consumer electronics. The final classification is based on the argument that the amount of service provided to customers of these goods is relatively minimal, especially compared to those who purchase and consume automobiles and personal computers. Both the automobiles and computers are, on average, more expensive purchases where service is more naturally bundled with the goods. However, the authors performed a sensitivity analysis and found that the results do not systematically change when appliances and electronics are classified as core goods with services.

Because the number of “core products with services” is relatively small (especially for the industry-level data), the authors begin with the goods vs. services classification. They then analyze the data using the four-level classification to provide additional insight.

Model Specifications

Reliability and customization are theoretically and conceptually distinct. Empirically, however, they are related. The authors’ approach to distinguishing between the impact of reliability and the impact of customization must take into account the fact that these measures are far from independent of each other. The ACSI model uses customization and reliability as redundant, reflective indicators of an overall quality construct. This presents problems when regressing both measures as independent variables on satisfaction. The authors’ solution is to analyze the impact of each measure separately and compare the results. If they find that customization and reliability behave very differently across conditions, it supports their use of separate analyses. If customization and reliability are so redundant that they behave exactly the same, hypothesis three will not be supported.

Both simple regression models and more general linear (analysis of variance) models were estimated separately for reliability and customization. The models were estimated for both firms (n = 823) and industries (n = 140) to provide a comprehensive test of the hypotheses. Although the authors focus on the firms, it is important to show consistency in the firm and industry results to rule out alternative explanations for the findings (such as fixed effects that exist for the firms but not the industries). The dependent variable in the models is customer satisfaction (the ACSI), while the independent variables are reliability or customization as continuous variables. The general linear models also include a product-to-service classification variable (two and four levels) and a five-level control variable to account for year-to-year variation in satisfaction. The models included the two-way interaction term involving reliability (or customization) and product-to-service classification.

To summarize, hypotheses one and two predict positive main effects for reliability and customization on satisfaction across firms and industries. Hypothesis three predicts that the effect of reliability increases relative to the effect of customization from goods to services. The interactions involving reliability/customization and the good (vs. service) classification are used to test hypothesis three. For example, the effect of reliability might decrease for goods (vs. services), while the effect of customization might increase. If both interactions are in the same direction, the interaction involving customization should at least be more positive for products than the interaction involving reliability.


Results of the simple regression models involving satisfaction, customization, and reliability for products vs. services are shown in Figure 1. The unstandardized betas or impact scores from these models reflect the change in the 0- to 100-point satisfaction index that results from a one-point increase in the 1- to 10-point customization or reliability scales.

The regressions show a greater effect of customization on satisfaction for products compared to services. The opposite pattern holds for reliability where the impact increases slightly for services. The general linear models and analysis of variance results for the firm-level sample reveal the significance of these differences. The models for customization and reliability explain 86 percent and 77 percent of the variation in customer satisfaction, respectively. Consistent with previous ACSI research (Fornell et al. 1996), both customization and reliability have a significant impact on satisfaction across goods and services (p < 0.001 for both customization and reliability). These results support hypotheses one and two.

The analysis of variance results also reveals a significant (p < 0.001) two-way interaction involving product vs. service classification and customization. That is, the impact of customization is significantly lower for services. In contrast, the interaction involving reliability was not significant. Directionally, the impact of reliability is higher for services. This confirms that the relative impact of reliability vs. customization increases as one moves from goods to service firms supporting hypothesis three. The industry-level models mirror what we find for the firms. The industry models for customization and reliability each explain about 80 percent of the variation in customer satisfaction. The analysis of variance results reveals highly significant main effects for both customization and reliability on satisfaction, as well as a significant interaction effect. The effect of customization on satisfaction decreases relative to the effect of reliability when moving from goods to services.

The four-level goods-to-services continuum provides greater insight into the results. The unstandardized betas or impact scores from the regression model results for the four-level classification are shown in Figure 2. Again, the impacts reflect the change in the 0- to 100-point satisfaction index that results from a one-point increase in the 1- to 10-point customization or reliability scales. The results show that the two service categories are very similar. In contrast, the greater impact of customization for goods is concentrated in the core goods category, while the lower impact of reliability for goods is concentrated in the pure goods category.

The general linear model results support the significance of these differences. The impacts of both customization and reliability are positive and significant in each of the four categories (p < 0.05). There is a product-to-service classification by customization interaction (p < 0.001), as well as a product-to-service by reliability interaction effect (p < 0.001). The pattern of effect sizes for the interactions is consistent with the previous results and provides some insight into what is driving them. Using the pure service classification as a baseline, the impact of customization on satisfaction is not significantly different for core services but higher for the pure goods category (p < 0.077) and much higher for the core goods category (p < .001). Again using pure services as a baseline, the effect of reliability on satisfaction is not significantly different across the pure service, core service, and core product categories. Reliability, however, is significantly less important for the pure goods category (p < 0.001).

Overall customization is more important for goods, where the effect is concentrated in the core goods category. Reliability is equally important in all but the pure goods category, where the impact is systematically lower. This finding is consistent with other research that finds customization to be much more important than reliability among the relatively “low tech” nondurable products that dominate the pure goods category (Johnson and Ettlie 2001). Given the relatively simple nature of nondurable products, their technology, and production, there are relatively few “things gone wrong.” Rather, the relative emphasis is on “things gone right” or customizing the goods to fit a variety of market segment needs.


This study and results show that customization (the provision of “things gone right”) and reliability (the reduction of “things gone wrong”) play different roles in driving customer satisfaction along the goods-to-services continuum. Reliability becomes relatively more important when compared to customization when moving from pure goods to pure services. The authors’ findings are consistent with the argument that coproduction makes reliability inherently more important for services (Grönroos 1990; Zeithaml, Parasuraman, and Berry 1996). The findings are inconsistent with arguments that customization is more important in a service context because of the one-to-one nature of service production.

This has important implications for where both goods and services should focus their quality improvement efforts. Customization is the more important quality driver across industries in this study. Clearly, however, service firms should focus relatively more attention on driving variation out of the production process. This finding is important in light of Sun’s (2001) research, which finds that service companies pay more attention to external customer satisfaction and loyalty, while manufacturing companies focus on quality assurance of internal manufacturing processes and products. The authors’ results suggest that it is time to shift resources such that service firms pay greater attention to improving basic reliability.

By categorizing firms along a goods-to-services continuum, the authors show that customization has its greatest impact among core goods, such as automobiles, where there is a significant service component. Reliability has its lowest impact among pure goods. As noted, this is consistent with the relatively simple and reliable nature of the production and distribution process for these products, which makes it difficult for reliability to have much impact on customer perceptions.

As Figures 1 and 2 illustrate, it is interesting that customization is the dominant driver of customer satisfaction across the goods-to-services continuum. One likely reason is that customization simply encompasses a wider array of the attributes and benefits that drive customer satisfaction (Johnson and Gustafsson 2000). Even in studies of service quality per se, reliability is but one of five benefit-level drivers of overall quality in the SERVQUAL methodology (Parasuraman, Zeithaml, and Berry 1988). The finding is important because it underscores the importance of meeting heterogeneous customer needs. As Fornell et al. (1996, 14) conclude, “squeezing more variance out of a manufacturing or service delivery process may not increase product quality or customer satisfaction as much as tailoring goods and services to meet customer and market segment needs.” However, the authors’ results do show that the impact of customization is significantly lower for service industries, while the impact of reliability is not significantly different between goods and services. Relatively, therefore, squeezing more variance out of service processes remains a critically important activity.

A strength of this study is that it is based on customers’ own perceptions of actual goods and services in the marketplace. Since quality is in the eyes of the beholder, customer perceptions are the best available measure of quality. At the same time, this creates a limitation. Although customization and reliability are conceptually distinct, they are empirically quite similar. This makes it difficult when using both customization and reliability as independent variables in the same analysis. The solution was simple—analyze them separately. If they are completely redundant perceptions of quality, then they should behave the same in this analysis. But clearly they did not. As a check, the authors also tested the hypotheses using structural equation models (following Gustafsson and Johnson 1997; Johnson and Ettlie 2001) where customization and reliability were treated as both reflective and formative indicators of latent quality as a driver of satisfaction. The results were completely consistent with those reported here. Whether using linear regression, analysis of variance, or structural equation modeling, these findings are robust.

The fact that the study shows reliability to be an important driver of customer satisfaction is itself an important contribution. Both scholars and practitioners have promoted customization heavily of late. It is critical not to ignore the role of reliability and quality assurance in the process. Consider, for example, a recent study by Curkovic, Vickery, and Droge (1999) in the automotive industry. These authors find that product reliability and durability have strong relationships to business performance. At the same time, they are low strategic priorities among the chief executive officers who participated in the study. To be able to retain and expand their customer base, an organization needs to implement new product attributes that correspond to customer needs. However, the organization must also make the product reliable. This requires a dual focus during product development, which incorporates the voice of the customer early in the process and subsequently breaks it down into different subsystems to assure reliability. One recent suggestion is to integrate customer satisfaction modeling with quality function deployment (Gustafsson and Johnson 1997). In latter phases of the development process, other methodologies for driving variation out of the production process include robust design and design of experiments.

To bring more clarity to the role that different quality dimensions have in driving customer satisfaction, future research might simultaneously collect both objective quality data and perceived quality. This would make it possible to gain a deeper understanding of the relationship between internal and external quality. Another appealing avenue is to follow a goods industry over time and study the changing role of customization and reliability. If the amount of services to goods in the product increases or changes over time, it would be informative to examine whether the impact of customization and reliability change as well. This would help researchers to better understand how a firm’s quality strategy should evolve over time.


The authors gratefully acknowledge the support of the National Quality Research Center at the University of Michigan Business School for providing the data used in the research.


Anderson, E. W., C. Fornell, and R. T. Rust. 1997. Customer satisfaction, productivity, and profitability: Differences between goods and services. Marketing Science 16 (spring): 129-145.

Bateson, J. E. G., and K. D. Hoffman. 1999. Managing services marketing: Text and readings. Fort Worth, Texas: Dryden Press.

Berry, L. L., A. Parasuraman, and V. A. Zeithaml. 1994. Improving service quality in America: Lessons learned. Academy of Management Executive 8, no. 2: 32-52.

Bolton, R. N. 1999. A dynamic model of the duration of the customer’s relationship with a continuous service provider: The role of satisfaction. Marketing Science 17, no. 1: 45-65.

Colgate, M. R., and P. J. Danaher. 2000. Implementing a customer relationship strategy: The asymmetric impact of poor versus excellent execution. Journal of the Academy of Marketing Science 28 (summer): 375-387.

Crosby, P. B. 1980. Quality is free. New York: Penguin Books.

Curkovic S., S. K. Vickery, and C. Droge. 1999. Quality and business performance: An empirical study of first-tier automotive suppliers. Quality Management Journal 6, no. 2: 29-40.

Deming, W. E. 1981. Management of statistical techniques for quality and productivity. New York: New York University.

Feigenbaum, A. V. 1991. Total quality control: Fortieth anniversary edition. New York: McGraw-Hill.

Fornell, C. 1992. A national customer satisfaction barometer: The Swedish experience. Journal of Marketing 56 (January): 6-21.

Fornell, C., and M. D. Johnson. 1993. Differentiation as a basis for explaining customer satisfaction across industries. Journal of Economic Psychology 14 (December): 681-696.

Fornell, C., M. D. Johnson, E. W. Anderson, J. Cha, and B. E. Bryant. 1996. The American customer satisfaction index: Nature, purpose and findings. Journal of Marketing 60 (October): 7-18

Garvin, D. A. 1984. What does ‘product quality’ really mean? Sloan Management Review 26 (fall): 25-43.

Gustafsson, A., and M. D. Johnson. 1997. Bridging the quality-satisfaction gap. Quality Management Journal 4, no. 4: 27-43.

Grönroos, C. 1990. Service management and marketing: Managing the moments of truth in service competition. Lexington, Mass.: Lexington Books.

Hair, J., R. E. Anderson, R. L. Tatham, and W. Black. 1995. Multivariate data analysis: With readings. London: Prentice Hall.

Hoffman, D. K., and J. E. G. Bateson. 1997. Essentials of service marketing. Fort Worth, Texas: Dryden Press.

Huff, L., C. Fornell, and E. W. Anderson. 1996. Quality and productivity: Contradictory and complementary. Quality Management Journal 4, no. 1: 22-39.

Ishikawa, K., and D. J. Lu. 1985. What is total quality control? The Japanese way. Englewood Cliffs, N.J: Prentice Hall.

Johnson, M. D., and J. E. Ettlie. 2001. Technology, customization and reliability. Journal of Quality Management 6: 193-210.

Johnson, M. D., and C. Fornell. 1991. A framework for comparing customer satisfaction across individuals and product categories. Journal of Economic Psychology 12 (June): 267-286.

Johnson, M. D., and A. Gustafsson. 2000. Improving customer satisfaction, loyalty and profit: An integrated measurement and management system. San Francisco, Calif.: Jossey-Bass.

Juran, J. M., and F. M. Gryna. 1988. Juran’s quality control handbook, fourth edition. New York: McGraw-Hill.

Kotler, P. 2000. Marketing management: The millennium edition. Upper Saddle River, N. J.: Prentice Hall.

Martin, C. R., and D. A. Horne. 1992. Restructuring toward a service orientation: The strategic challenges. International Journal of Service Industry Management 3, no. 1: 25-38.

Mittal, V., P. Kumar, and M. Tsiros. 1999. Attribute-level performance, satisfaction, and behavioral intentions over time: A consumption-system approach. Journal of Marketing 63 (April): 88-101.

Parasuraman, A., L. L. Berry, and V. A. Zeithaml. 1991a. Understanding customer expectations of service. Sloan Management Review 32 (spring): 41-50.

– – –. 1991b. Understanding, measuring, and improving service quality: Findings from a multiphase research program. In Service Quality: Multidisciplinary and Multinational Perspectives, edited by S. W. Brown, E. Gummesson, B. Edvardson, and B. Gustavsson. New York: Lexington Books.

Parasuraman, A., V. A. Zeithaml, and L. L. Berry. 1985. A conceptual model of service quality and its implications for future research. Journal of Marketing 49 (fall): 41-50.

– – –. 1988. SERVQUAL: A multiple-item scale for measuring consumer perceptions of service quality. Journal of Retailing 64 (spring): 12-40.

Reeves, C. A., and D. A. Bednar. 1994. Defining quality: Alternatives and implications. Academy of Management Review 19, no. 3: 419-445.

Smith, A. K., R. N. Bolton, and J. Wagner. 1999. A model of customer satisfaction with service encounters involving failure and recovery. Journal of Marketing 36 (August): 356-372.

Sun, H. 2001. Comparing quality management practices in the manufacturing and service industries: Learning opportunities. Quality Management Journal 8, no. 2: 53-71.

Zeithaml, V. A., A. Parasuraman, and L. L. Berry. 1990. Delivering service quality. New York: The Free Press.

– – –. 1996. The behavioral consequences of service quality. Journal of Marketing 60 (April): 31-46.


Michael D. Johnson is the D. Maynard Phelps Collegiate Professor of Business Administration and professor of marketing at the University of Michigan Business School. He is the author of four books, including Customer Orientation and Market Action (Prentice Hall, 1998) and Improving Customer Satisfaction, Loyalty and Profit (Jossey-Bass, 2000), as well as more than 100 academic articles and reports. Johnson has been instrumental in the development of the Swedish Customer Satisfaction Barometer (SCSB), the American Customer Satisfaction Index (ACSI), and the Norwegian Customer Satisfaction Barometer (NCSB). He currently works with a variety of companies and public agencies on issues pertaining to satisfaction and loyalty measurement, relationship management, and quality improvement.

Johnson earned his doctorate in behavioral science and marketing from the University of Chicago. He may be contacted at the University of Michigan Business School, 701 Tappan Street, Ann Arbor, MI 48109-1234; 734-764-1259; fax: 734-963-0274; e-mail: .

Lars Nilsson is assistant professor of business economics in the Service Research Center at Karlstad University, Sweden. His research on quality practices, continuous improvement, and product development has been published in the Quality Management Journal, the Journal of Quality Management, Total Quality Management, and the International Journal of Technology Management.

Nilsson earned his doctorate in quality technology and management from Linköping University, Sweden. His dissertation is titled Quality Practice: An Empirical Investigation of Product Development and the Goods-to-Services Continuum. He was a visiting scholar at the University of Michigan Business School during the summer and autumn of 1999. He may be contacted at the Service Research Center, Karlstad University, Karlstad SE-651 88, Sweden; 54-700-21-34; fax: 54-83-65-52; e-mail: .


Return to top

Featured advertisers


(0) Member Reviews

Featured advertisers

ASQ is a global community of people passionate about quality, who use the tools, their ideas and expertise to make our world work better. ASQ: The Global Voice of Quality.