In this paper we outline an overall approach for tracking service and relationship quality that differs substantially from the conventional paradigm. The conventional attitude-based approach relies on assumptions about the link between evaluations of service quality and subsequent behaviour which are not supported by the substantive body of research findings about buyer behaviour. The attitude-based approach also requires inferences to be made concerning what aspects of service provision determine the attitudes. Unfortunately the traditional approach is still widely practiced in commercial market research, as well as practised and taught in Universities. We therefore now explain a different approach which looks directly at service provision by using descriptive non-attitudinal measures and seeks to relate these to observed buying behaviours of interest. The approach has been trialled across a number of service categories and markets with promising results. Introduction Most service companies have research programs designed to measure service quality, and/or customer satisfaction, and/or relationship quality. Such programs are designed to allow management to manage service provision and relationship building initiatives. They provide essential information to guide efforts to reduce variability in service quality and to provide customers with the service that will help ensure their continued patronage. While there is little direct evidence as to the link between service quality and better company performance, company-level data suggests a link between higher quality, higher market share and improved profitability (Buzzell and Gale 1987; Buzzell et al. 1975; Rust and Zahorik 1993). The Problem with Subjective Evaluations/Attitudes Measuring service quality and satisfaction traditionally involves asking customers for subjective attitudinal evaluations, that is, asking if they personally felt the service they received was satisfactory. Many service and relationship quality measurement programs also ask customers for subjective evaluations, eg "please rate the quality of the delivery service was it excellent, very good, fair or poor". This is done on the assumption that customers' attitudes towards the service will affect their future behaviour, such as repeat-purchase behaviour or recommending the company to others. Attitude à future behaviour (buying or recommending) While this attitudinal measurement approach sounds intuitively sensible, it is actually fraught with problems. It is a needlessly complicated and indirect approach. It does not focus on service provision or buyer behaviour. Instead it measures customer attitudes and makes inferences from these both to service delivery and to buyer behaviour. Yet the purpose of service quality market research is not to determine whether or not customers give positive or negative evaluations (actually the answer to this question is usually the same most customers give positive evaluations). What really needs to be discovered is (a) what sort of service is the company really delivering, and (b) how does this impact on real customer behaviour, that is, what aspects of the service really affect behaviour either positively and negatively ? So logically, the important things are to measure the actual service provision experienced and the subsequent behaviour. Service delivery à future behaviour (buying or recommending) Apart from this logical argument against measuring customers subjective evaluations (attitudes) there are also three very practical concerns: Attitude-based service quality research does not tend to produce managerially actionable findings. Managers need to know what specific aspects of service delivery are poor and which aspects affect behaviour. Measuring global attitudes is not a direct way of gathering this information. Finding out that on average, customers rate the hotel's reception service at "6 out of 10" does not tell a manager what the problems with the reception service actually is. ANZMAC 2000 Visionary Marketing for the 21 Century: Facing the Challenge 1136 New research into customer's subjective evaluations (attitudes) has shown that customers are extremely fickle. Typically, only half the respondents who gave a particular attitudinal response (eg "Ansett has good in-flight service") give the same response on a second interview, irrespective of when that second interview takes place for example 15 minutes later or a month later (Dall'Olmo Riley et al. 1997). It seems that attitudes are seldom deeply held, customers give an attitudinal response only because they were asked to, not because they necessarily hold a strong opinion. Most of the time customers form no real judgement about the service, and even when they do they often quickly forget it. This is evidenced by the wide scale success of the Dirichlet model of repeat-purchase that assumes no purchase feedback (Goodhardt et al. 1984). Not surprisingly then, decades of research into the link between customer's expressed attitudes and their subsequent behaviour has consistently shown a very weak link (Foxall 1999; Kraus 1995). The vast bulk of customers who switch away from the brand they used previously said they were satisfied. Attitudes appear to be more a function of behaviour (we like brands we have used) rather than the other way round (Bird and Ehrenberg 1966). Therefore, basing service quality research on attitudes is not sensible or useful. Likewise there seems little point in measuring customer satisfaction (which is nothing more than an attitude). The exceptions to this are organisations in a monopoly position, such companies do not need to worry about customer loyalty but they usually need to keep customers satisfied in order to avoid regulators or at least fulfil the requirements that regulators have set them, for example Telstra's rural obligations. Also, monopoly companies can not measure loyal behaviour all their customers have to be loyal. Companies with government ownership, and government organisations themselves are also rightly concerned with customer (voter) satisfaction. But most companies are not in monopoly positions so their emphasis should be on measuring their service provision and relationship enhancing efforts, and determining how these impact on buying behaviour. In this paper we briefly describe a new approach based on the above rationale. Linking Service Provision with Buyer Behaviour Given the problems with evaluative (attitudinal) assessments of service quality our approach emphasises the gathering of descriptive, 'objective', information on the service provision and on customer behaviour (either repeat-purchase or providing recommendations). Making causal inferences about objective elements of service provision, even from cross sectional data, is often less contentious than it is from attitudinal variables. There is clearer separation between the independent variables (service that did or did not occur) and the dependent variable (loyalty behaviour). Whereas the separation between say satisfaction and re-purchase intention is not great. Even between attitudes and behaviour it is difficult to know what is causing what. Objective measures are not only possible, but they form the basis of much scientific work even in the social sciences (see Hunt 1993). The use of objective measures of service quality are, however, uncommon (though there is some use in commercial market research e.g., mystery shopping, and much use in product quality research). A reason for this reluctance to employ measures that are based on pre-determined explicit standards may be that such measures are seen as not customer-oriented. That is, that they may not capture customers' assessments of the service experience. This can be addressed by prior qualitative research to help ensure that a service quality study captures relevant aspects of the service received. Another important benefit of a prior qualitative study is to ensure that the instrument uses customers’ own terminology to increase the validity of the study (to make sure all respondents understand the questions and to remove any ambiguous questions or jargon). To collect descriptive information on the service provision we may: ask customers to provide descriptive responses concerning the actual service they received (e.g., "did the porter offer to take your bags ?, "were you given street directions ?"); or have trained 'mystery shoppers' record the service that was delivered. Mystery shopping has been found to be a reliable and valid means of evaluating service quality, although the results cannot be used to compare individual sites due to service variability and sampling concerns (Dawes and Sharp 2000); or ask staff to factually report what they did (or did not do) ; or use internal company records to provide data on what relationship building interventions the customer was and was not exposed to. ANZMAC 2000 Visionary Marketing for the 21 Century: Facing the Challenge 1137 Ideally the market research should explicitly measure the effectiveness of organisations’ selling channels (sales representatives, telephone sales etc) and also customers’ evaluations of the service received in ordering, gaining and using the product or service. For example, in the airline industry a customer service quality study would incorporate measures for the purchasing of tickets, pre-flight service (check-in, baggage handling, lounge), the actual flight itself (leaving on time, food service, special requests, general presentation and hospitality of crew), post flight service (baggage, connecting flights etc) and any follow-up contact (eg. complaint handling). This information is can then be modelled against data on customer behaviour that is also collected. Behaviours such as repeat-buying, recommending, complaining, or paying price premiums. Where it isn't possible to measure actual behaviour we advocate the use of non-attitudinal, probabilistic predictor

Service quality and satisfaction have been, and will continue to be, key constructs of interest to marketers. Scepticism about the importance to an organisation of service quality makes it imperative that research be undertaken to address the quantification of the impact of marketing interventions intended to result in increased levels of loyalty among customers (Rust and Zahorik, 1993). However, to date, multiple empirical replications demonstrating the magnitude of the relationship between service quality, customer satisfaction and consumers' loyalty has been lacking. Traditionally, explanations of consumers' services purchases have focussed on the relationship between service quality and purchase intentions (Rust and Zahorik, 1993, Parasuraman et al., 1991, Parasuraman et al., 1985, Cronin and Taylor, 1994, Parasuraman et al., 1994). This poses a dilemma for management because it is also well known that consumers do not always buy the highest quality service (Olshavsky, 1994).

Competitive brands within a product category seldom differ much in overall customer perceived service quality. We report descriptive statistics (providing grounded benchmarks) for service quality. While we show that there are differences in average service quality scores between industries, competitive brands within an industry show almost no difference in mean service quality scores. Likewise their variation in individual scores is typically identical. This pattern, which runs completely counter to that assumed by the service quality literature, was found to hold across six industries, 56 brands, more than 30 countries and is based on in excess of 89,000 customer interviews using a wide range of survey and measurement approaches.