1
International Journal of Market Research, Volume 55, Issue 2, Pages 227-246
Towards a Better Measure of Customer Experience
Philipp Klaus and Stan Maklan
Defining and improving customer experience is a growing priority for market
research because experience is replacing quality as the competitive battleground for
marketing. Service quality, we argue, is an outgrowth of the Total Quality
Management movement of the 1980s and suffers from that movement’s focus on the
provider rather than the value derived by customers. The most popular measure of
service quality – SERVQUAL – assesses the functional delivery of service during a
single episode. This conceptualisation allows service to be improved along traditional
quality management principles. The increasingly settled view of researchers is that
customer experience is generated through a longer process of company-customer
interaction across multiple channels and is generated through both functional and
emotional clues. Research with practitioners indicates that most firms use customer
satisfaction, or its derivative Net Promoter Score, to assess their customers’
experiences. We question this practice based on the conceptual gap between these
measures and the customer experience. In the IJMR October 2011, we proposed the
principles of a new measure appropriate for the modern conceptualisation of customer
experience: the Customer Experience Quality (EXQ) scale. In this article we extend
that work to four service contexts to support a claim of generalisability better and
compare its predictive power with that of customer satisfaction. We propose that EXQ
better explains behavioural intention and recommendation than customer satisfaction.
The background
In our article asking if market researchers were using the right measures to help their
firms improve customer experience, we established that customer experience was
conceptually different from service quality and hence requires a new corresponding
measurement (Klaus & Maklan, 2007). The role of measurement in successfully
implementing and executing strategy is long established and well documented (e.g.
Martilla & James 1977). This role is particularly crucial for new emerging paradigm
shifts (Bowden 2009) such as the most recent one towards customer experience
management (Smith 2002).
Based on research and literature, we defined customer experience as: the customer’s
cognitive and affective assessment of all direct and indirect encounters with the firm
relating to their purchasing behaviour. This definition is highly consistent with
conceptualisations offered by other researchers (e.g. Lemke et al. 2010, Verhoef et al.
2009).
Our interpretation of the literature generated the core tenants of a measure for
customer experience (Maklan & Klaus 2011):
1. It is assessed as an overall perception by customers and not as a gap to
expectations.
2
2. Customers’ assessment is based on overall value in use and not just a
summation of performance during individual service episodes.
3. The measure of experience has a broader scope than that proposed by
SERVQUAL. It includes emotions and peer influences.
4. Experience begins before service encounters and continues after the
encounters.
5. Experience is assessed against service encounters across all channels.
6. An ideal measure should link more directly to customer behaviour and
business performance than do either SERVQUAL or customer satisfaction.
We presented EXQ, a scale measure of customer that explains important marketing
outcomes (Maklan & Klaus 2011). We illustrated that it incorporates key attributes of
customer experience that are not captured in current market research assessments of
service quality or customer satisfaction. EXQ allows tracking both of customer
experience and its key attributes over time and can act as an important marketing
metric.
Readers and managers commenting upon the article ask two main questions: (1) is
EXQ generalisable beyond the context we presented, and (2) does it add to our current
research practice based largely upon customer satisfaction and its derivative Net
Promoter Score? This paper presents a synthesis of four contexts in which we
validated EXQ to address issues of generalisability and assesses EXQ’s explanatory
power against customer satisfaction.
The rest of the paper is structured as follows. Next we respond to readers and
managers by developing the hypotheses that address their questions. Then we
introduce the method to address these aims. Finally we present and discuss our
findings, including research contributions, managerial implications and suggestions
for future research.
Hypotheses development
This paper focuses on the impact of customer experience on important marketing
outcomes, namely customer satisfaction, loyalty and word-of-mouth behaviour. We
chose these outcomes based on the weight of research suggesting their importance as
outcomes (e.g. Puccinelli et al. 2009).
Scholars posit customer experience as a key determinant of customer satisfaction and
loyalty (e.g. Caruana 2002). Customer experience and satisfaction, while discrete
constructs (Garbarino & Johnson 1999), are connected through a contributory
relationship (Fornell 1992). Researchers suggest experience drives satisfaction, which
in turn drives loyalty (e.g. Shankar et al. 2003). Marketing scholars acknowledge the
link between satisfaction and loyalty intentions (Yi & La 2004). The exact nature of
this relationship is still questioned because improved customer satisfaction is a
desirable, but not sufficient, condition for affecting behavioural intentions
(McDougall & Levesque 2000). Therefore we explore the influence of customer
experience on customer satisfaction and loyalty intentions independently.
Customer experience not only drives customer satisfaction (e.g. Anderson & Mittal
3
2000) and loyalty (Fornell et al. 2006), but also word-of-mouth (Keiningham et al.
2007). The influence of customer experience on word-of-mouth (recommendation) is
discussed widely in traditional offline media (Babin et al., 2005), online (e.g. Hennig-
Thurau et al. 2002) and experiential settings (e.g. Voss & Zomerdijk 2007).
Subsequently, we explore the proposed relationship between customer experience and
word-of-mouth.
We believe, based on our synthesis of the literature, that experience has a more
significant impact than satisfaction on customer loyalty and word-of-mouth (e.g.
Koenig-Lewis & Palmer 2008). That customer satisfaction has a positive influence on
behavioural loyalty intentions (e.g. Gremler & Brown 1996), and between customer
satisfaction and word-of-mouth (e.g. Brown et al. 2005) has been researched widely.
Therefore we refrain from stating hypotheses between these constructs.
Based on literature, we developed the following hypotheses (see Figure 1):
Hypothesis 1: Customer experience has a significant positive impact on customer
satisfaction.
Hypothesis 2: Customer experience has a significant positive impact on loyalty
intentions
1
.
Hypothesis 3: Customer experience has a significant positive impact on word-of-
mouth behaviour.
Hypothesis 4: Customer experience has a more significant positive impact on loyalty
intentions than customer satisfaction.
Hypothesis 5: Customer experience has a more significant positive impact on word-
of-mouth behaviour than customer satisfaction.
In this paper, we validated a customer experience quality scale (EXQ) that can be
readily adapted by different types of service providers. In order to develop and
validate a customer experience quality scale capable of serving this purpose, we
adopted and extended the validated Silvestro et al. (1992) service classification
scheme. We chose one high value service (mortgages), one mass service (fuel and
service station) and one “utility” service (retail banking). In addition, we included one
service reflecting the hedonic nature of customer experiences: lifestyle luxury goods
retail. The latter service was chosen to ensure further cross-validation (Cronin et al.
2000), so that samples varied on the degree to which the service could be
characterised as hedonic (lifestyle luxury goods) versus utilitarian (fuel and service
station). The context for the research was the UK, although most of the firms operate
internationally.
We adopted Klaus and Maklan’s (2012) Customer Experience Quality (EXQ) scale to
validate our hypotheses about the impact of customer experience quality and its
impact on important marketing outcomes, namely customer satisfaction, loyalty and
word-of-mouth behaviour, as suggested and defined by Brown et al. (2005), Dagger
et al. (2007), Parasuraman et al. (2005), Walsh and Beatty (2007) and Zeithaml et al.
(1996). A full description of the corresponding measures is provided in Appendix 1.
1
As defined and measured by Zeithaml et al. 1996
4
In the 2011 article, we introduced a scale measure of customer experience - EXQ
whose dimensions are product experience, outcome focus, moments-of-truth and
peace-of-mind (POMP). Product experience refers to the importance of customers’
perception of having choices and the ability to compare offerings, a critical factor in
modelling consumer behaviour (McAlister & Srivastava 1991) and antecedent of
loyalty (Srinivasan et al. 1998). Outcome focus is associated with reducing customers’
transaction cost, such as seeking out and qualifying new providers, reflecting the
importance of goal-oriented experiences in consumer behaviour (Huffman & Houston
1993). Moments-of-truth emphasizes the importance of service recovery (Tax et al.
1998) and flexibility (Liljander & Strandvik 1993) in dealing with customers once
complications arise. Peace-of-mind describes the customer’s assessment of all the
interactions with the service provider before, during and after the purchase of the
service. This dimension includes statements strongly associated with the emotional
aspects of service (Klaus & Maklan 2011; Edvardsson 2005). The full list of attributes
for each dimension is provided in Appendix 1. The authors are happy to share further
data of method, questionnaire, results and attendant validity testing with interested
readers.
We collected the data as follows:
Mortgage customers: an online questionnaire accessible through a link sent by a
market research firm to a sample of customers of a UK bank who had
purchased their most recent mortgage within the previous six months.
Fuel and service station customers: an online questionnaire accessible through a
link sent by a market research firm to a sample of customers in their customer
database, split between first time and repeat customers.
Retail banking customers: a printed questionnaire, distributed to customers of
the retail bank, split between first time and repeat customers.
Luxury goods customers: an online questionnaire accessible through a link sent
by a market research firm to a sample of customers who had purchased items
with the service provider within the previous three months, split between first
time and repeat customers.
Respondents rated their customer experience on each scale item using a 7-point scale
(1 = Strongly disagree, 7 = Strongly agree) or as Do not know/Not applicable. The
items were grouped by dimensions for expositional convenience; they appeared in
random order on the survey. The symbols preceding the items correspond to the
variables named in Figure 1 (see Appendix 1). The corresponding survey generated
800 qualified responses (200 responses per context), which were subsequently
analysed utilising the software packages SPSS 16.0 and AMOS 16.0. Appendix 2
contains descriptive profiles of the exploratory stage of each context. Before running
the structural model, we examined whether the four samples could be pooled or
demanded three separate analyses. The results of the multigroup comparison
confirmed configural invariance (CFI 0.97; RMSEA 0.05) and factor loading
equivalence (CFI 0.97; RMSEA 0.05; with an insignificant change in chi-square of
8.9/df 997). These values indicate metric invariance, which implies that the three
samples represent the same general population (Hair et al. 1998). Therefore, we
proceeded with an analysis based on pooled data (see Table 1). The response bias
analysis (Armstrong & Overton 1977) provided evidence that non-response was not a
concern in this study. Managers in the individual companies reviewed the
5
demographic profiles of the respondent samples and confirmed that they were
representative of their customer base.
The fit of the measurement and structural models examined was assessed through
multiple indices, as recommended by Hoyle and Panter (1995). It has been suggested
that a Chi-square value two or three times as large as the degrees of freedom is
acceptable (Carmines & McIver 1981), but the fit is considered better the closer the
Chi-square value (CMIN) is to the degrees of freedom (df) for a model (Thacker et al.
1989). EXQ’s CMIN/df ratio displays an excellent fit. Measures of incremental fit
were used as indicators of acceptable model fit. The scale statistics (see Table 2)
indicate the robustness of the EXQ model (Garver & Mentzer 1999; Hoyle and Panter
1995) on the basis of the fit criteria established in prior service quality research
(Parasuraman et al. 2005).
The psychometric properties of the scale were evaluated through a comprehensive
Confirmatory Factor Analysis (CFA). All items were tested in the same model and
were restricted to load on their respective factors. The results are a sign of high levels
of construct reliability and average variance extracted for all latent variables. All t
values were significant (p = 0.05), and the average variances extracted were >0.50,
and thus convergent validity was established. Using Fornell’s and Larcker’s (1981)
stringent criteria for measuring the internal consistency of a scale and its ability to
measure a latent construct, we establish construct reliability with estimates exceeding
0.50 (see Table 3).
After establishing the strength and psychometric properties of the scales underpinning
the model, we examined the structure of the model. We modelled customer
experience as suggested by researchers as a formative construct in which the
dimensions of the model drive customer experience perceptions (Parasuraman et al.
2005). In addition, we conducted second-order CFAs in which the dimensions of
EXQ (e.g. peace-of-mind) were specified as reflective indicators of a second-order
overall customer experience (EXQ) construct. The CFA analysis and model fit
statistics were analogous to those reported in this study. On the basis of these criteria,
we modelled our measurement reflectively (see Figure 1). Therefore, the confirmatory
factor analysis (CFA) results reported are for first-order factor models specifying the
scale items as reflective indicators of their corresponding latent constructs, and allow
the latent constructs to intercorrelate.
To establish nomological validity, we examined how well the EXQ scale relates to
other variables. Thus, in addition to the EXQ scale, the questionnaire included a five-
item Behavioural Loyalty Scale (Parasuraman et al. 2005) based on a 13-item battery
developed by Zeithaml et al. (1996); adapted a 5-item Customer Satisfaction scale
(Dagger et al. 2007); and incorporated a 7-item Word-of-Mouth Behaviour scale
(Brown et al. 2005). These measures (see Appendix B) allow us to capture the full
range of potential behaviours likely to be triggered by customer experience
(Mascarenhas et al. 2006). To demonstrate that a measure has nomological validity,
the correlation between the measure and other related constructs should behave as
expected in theory (Churchill 1995). The reliability of these scales was assessed with
a composite reliability coefficient (ranging from 0.92 to 0.97) and CFA, which clearly
confirmed the appropriateness of the operationalisations (see Table 4).
