Journal of Retailing and Consumer Services 62 (2021) 102661

Contents lists available at ScienceDirect

Journal of Retailing and Consumer Services

journal homepage: www.elsevier.com/locate/jretconser

Robotic service quality – Scale development and validation

Catherine Prentice a, *, Mai Nguyen b
a
Department of Marketing Griffith Institute for Tourism Griffith Business School, Griffith University Staffordshire Business School Staffordshire University, Australia
b
Department of Marketing, Griffith Business School Griffith University | Nathan Campus | QLD 4111, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Viewing robots as service agents that provide services to customers for value exchange, the study developed a
Robots scale to measure robotic service quality. The scale underwent several stages of development including item
Artificial intelligence generation, domain specification, scale refinement, and validity testing, including internal and external cross
Service quality
validation. A range of methods were used in this process. Data were collected from Australia, China, and Vietnam
Scale development
Services marketing
to test external validity. Four dimensions were identified to represent robotic service quality. Development of this
scale has implications for artificial intelligence and service research. The scale can be used by practitioners to
enhance customer experience and generate positive attitudinal and behavioural responses from customers.

1. Introduction assessment. This paper views service robots as service providers that
deliver a service to customers to generate positive attitudes and be­
Robots with advanced AI technology and machine learning have haviours towards the service firm for value creation. Consistent with this
been utilised in the service sector to facilitate business operations and view, the aim of this study is to develop a scale to measure robotic
customer service. Gartner (2019) suggested that 37 percent of organi­ services, referred to as robotic service quality. Given the prevalent
sations had adopted robots and AI in some form in 2019, increasing by application of robotic use in customer service, the development of this
270 percent over the past four years. Robotic service today has been scale would help service firms to assess the role of robotic service in the
integrated into service organisations to enhance customer experience organisation and the potential influence on customers’ attitudes and
(Prentice et al., 2020). For example, Chatbots have been deployed to behaviours towards the entity.
enable self-service procedures (Kumar and Balaramachandran, 2018). Service quality in the marketing literature refers to customers’ per­
The concierge robot “Connie” (e.g. Hilton Worldwide) can interact with ceptions and value-judgments of a product or service. Service quality has
hotel guests and respond to individualised questions. The Henn-na Hotel also been recognised as an imperative factor for competitive advantage
in Japan uses robots for frontline and concierge services, as well as and organisational profitability (Shi et al., 2014). These outcomes result
assisting check-in and check-out (Rajesh, 2015). These service robots from positive customer responses manifested in market share expansion,
function as service employees that deliver customer service as well as customer satisfaction, and loyalty (Ozment and Morash, 1994). The
facilitate employees performance of service tasks (Cobos et al., 2016; underlying logic is that improving service quality leads to customer
Marinova et al., 2017; West et al., 2018). satisfaction and customer loyalty, which relates to business profitability
Wirtz and Lovelock (2018) discussed the role of service robots in the (Heskett et al., 1994). Consequently, measuring a firm’s service quality
service industry and compared their features and capabilities with those is imperative to understand its impact on organisational outcomes.
of frontline employees on micro, meso, and macro levels. The authors The scale development in this study draws upon the conceptualisa­
indicated that service robots can replace human employees for some tions of service and service quality. This initiative provides a better
low-level service tasks, as well as collaborate with employees to achieve understanding of robotic service and its impact on organisational out­
high-level service and facilitate service delivery. The quality of customer comes as service quality has been widely acknowledged as an ante­
service is determined by customers’ perceptions and evaluations (Par­ cedent of customer satisfaction and loyalty. The paper first discusses
asuraman et al., 1994). Service quality assessment has been well docu­ robotic service as a product that a service organisation provides to
mented although robotic service is relatively new to the literature. No customers for value creation (e.g. customer satisfaction, purchase, and
research to date has incorporated robotic service within service quality loyalty behaviours). Value is manifested in the level of robotic service

* Corresponding author. Griffith University Department of marketing Griffith Institute for Tourism Nathan, Queensland 4111 Australia.
E-mail addresses: [email protected], [email protected] (C. Prentice), [email protected] (M. Nguyen).

https://doi.org/10.1016/j.jretconser.2021.102661
Received 27 April 2021; Received in revised form 8 June 2021; Accepted 28 June 2021
Available online 6 July 2021
0969-6989/© 2021 Elsevier Ltd. All rights reserved.
C. Prentice and M. Nguyen Journal of Retailing and Consumer Services 62 (2021) 102661

quality assessed by customers. The existing service quality models will needs and preferences for tailored service provision. In the retail in­
be briefly discussed to determine the protocols that robotic service dustry, “Watson” is used by the North Face to support customers to
quality measurement draw on. The scale development process will then choose the most suitable jackets by providing personalised recommen­
be elaborated. The results and implications conclude this paper. dations based on an analysis of a large multivariance dataset. Robots
provide quick and accurate personalised suggestions to enhance
2. Literature review customer-brand interaction and also save on human labour costs within
the organisation (Gursoy et al., 2019). These technical outcomes are
2.1. Robotic service reflective of service performance, acts, or efforts that are intentionally
developed to increase value creation.
Robots can be described as intelligent physical devices with pro­
grammed autonomy, mobility, and sensor ability to perform certain 2.2. Robotic service quality
human tasks, or intelligent machines acting like humans to perceive,
learn, memorise, reason, and solve problems through machine learning, Any service that is offered for value exchange must be measured to
deep learning, or natural language processing (Russell et al., 2016). assess its quality. In the marketing literature, the value of service is
Robots can interpret and learn from external data for predefined goals assessed based on the quality level and its impact on customers as any
and tasks (Kaplan and Haenlein, 2019). Robots in the service sector, form of service should aim to generate positive customer responses and
referred to as service robots, can have a physical (e.g. Pepper) or virtual business profitability (Prentice, 2019; Zeithaml, 2000). In the services
(e.g. Alexa) representation, with a humanoid (e.g. Sophia) or marketing literature, service quality is defined as the customer’s
non-humanoid appearance (e.g. Roomba cleaning robot) (Wirtz et al., perception or evaluation of a service organisation’s overall excellence or
2018). The physical humanoid and non-humanoid robots can function superiority (Parasuraman et al., 1988). Empirical research has examined
as frontline service employees to deliver customer service, and virtual and confirmed a link between service quality, customer satisfaction, and
robots can operate as e-service providers (Wirtz et al., 2018). Consistent loyalty (Helgesen, 2006; Kim, 2011; Woodside et al., 1989; Zeithaml,
with this analogy, the former are referred to as frontline service robots, 2000). A high level of service quality leads to customer satisfaction and
and the latter as e-service robots. loyalty behaviours (Shi et al., 2014). Hence, assessing and evaluating
Service in the marketing literature is defined as an act, deed, or service quality is imperative for an understanding of its impact on
performance that is offered as a value-added activity to benefit stake­ organisational outcomes.
holders (Wirtz and Lovelock, 2018). Services are generally categorised Viewing robotic services as one type of customer service, this paper
into four domains: people processing, possession processing, mental argues that the value must be reflective of the level of quality based on
stimulus processing, and information processing (Wirtz and Lovelock, users’ perceptions and evaluations. Various service quality measures
2018). People processing services refer to those that directly serve cus­ and models have emerged over last three decades across different con­
tomers, such as medical centres providing medical services and the ex­ texts, for instance, retail (Dabholkar et al., 1996; Siu & Tak-Hing
amination of patients, requiring simultaneous participation from both Cheung, 2001), education (Brochado, 2009), health (Akter et al.,
the service provider and the customer. Robots can provide such services. 2013), casinos (Wong and Fong, 2012), and online services (Boshoff,
For instance, robotic medical assistants have been used to monitor pa­ 2007; Ding et al., 2011; Yang et al., 2004). Of these, models proposed by
tients and alert the medical staff when needed. Possession processing Gronroos (1982), Parasuraman et al. (1988), Parasuraman et al. (1985),
services refer to services directed at customers’ tangible possessions, for and Brady and Cronin (2001) have attracted the most attention and have
example, house cleaning. A robotic vacuum cleaner can perform this been extensively cited. Gronroos (date) assessed service quality from the
kind of service. Mental stimulus processing services refer to offerings technical and functional perspectives. The former refers to the outcome
that serve in a mental or educational capacity. Robots with big data of the service performance or what the customer receives within the
analytics are used to create personalised learning experiences, tailored service encounter. The latter indicates the subjective perception of ser­
to students’ individual abilities and needs (Rouhiainen, 2019). Social vice delivery. Parasuraman et al.’s (1988) SERVQUAL describes service
robots are used to create personalised learning experiences tailoring to quality as the discrepancy between customers’ expectations and per­
individual students’ abilities and needs (Belpaeme et al., 2018). Infor­ ceptions of a service. Cronin and Taylor (1992) argued that adding
mation processing services refer to information-based offerings, such as customer expectation may be inefficient and unnecessary, and cus­
banking, accounting, and legal services. AI-powered chatbots provide tomers’ perceptions alone are adequate to capture an assessment of the
round-the-clock information-based services, using advanced data ana­ overall service quality (McDougall and Levesque, 1995). The rapid
lytics for fraudulent transactions and compliance improvement. development of e-commerce has prompted numerous e-service quality
Robotic service is functioned through machine learning from the scales in the literature. For instance, an e-service quality scale by Santos
environment and past experience to improve decision making (Wu and (2003), e-retailing service quality by Lee and Lin (2005), and Collier and
Tegmark, 2019). It may also function through deep learning by Bienstock (2006), and an e-travel service quality scale by Ho and Lee
mimicking the human brain to analyse data and draw conclusions (2007). These scales are reflective of the quality of service delivered in a
manifested in image recognition, sound recognition, recommender virtual space.
systems, and natural language processing (Wu and Tegmark, 2019). Despite discrepancies in these measurements, these models consist of
However, the level of functional quality is manifested in technical out­ tangible and intangible services delivered through personal and imper­
comes by providing a consistent and timely service (West et al., 2018). sonal encounters. A service encounter is dyadic with limited or narrow
For instance, robots in hotels are deployed to carry out concierge and relational contact and communication (Grönroos, 1994). Personal en­
room services and assist with check-in and check-out, order a taxi, counters generally refer to interpersonal interactions between service
answer customers’ queries, and to refine responses based on customers’ employees who deliver the service and customers who receive the ser­
behaviour and feedback (Bartneck et al., 2009; Belanche et al., 2020). vice (Inanov et al., 2018; Prentice, 2019). Impersonal encounters indi­
Hilton Worldwide employs a robotic concierge named “Connie” to cate, that in the absence of employee service, customers’ interact with
interact with customers just as a regular frontline employee. “Connie” the physical or cyber/virtual/online settings of the service organisation
has the capacity to personalise a customer’s experience, by providing (Chuah and Yu, 2021; Prentice, 2019; Yang et al., 2004). The physical
information and addressing general needs (Tavakoli and Mura, 2018). In settings include any tangible offerings, servicescape, ambience, or at­
the airline industry, KLM Airlines has adopted “Spencer” to answer mosphere. Online settings can be any virtual encounters.
travellers’ queries and enhance their travel experience (West et al., Consistent with the foregoing discussion, a service quality measure
2018). The 1 A-TA robot is used by travel agencies to identify travellers’ should focus on customers’ perceptions and evaluations of their

2
C. Prentice and M. Nguyen Journal of Retailing and Consumer Services 62 (2021) 102661

encounters with service providers. Hence, robotic service quality should invited to assess the importance of these items using a seven-point scale
also be assessed by these encounters with customers and assessment ranging from 1 “not important” to 7 “very important”. Any item with an
must be drawn from customers perceptions and evaluations. The tech­ average score less than 5 or “somewhat important” were removed from
nology acceptance literature indicates that technology quality must be the study (Brakus et al., 2009). This process resulted in 18 items for scale
assessed from an information and systems perspective and evaluate how refinement and validation.
these factors affect users’ beliefs, attitudes, and behaviours (Wixom and
Todd, 2005; Xu et al., 2017). Consistent with this view, this study will 3.2. Stage 2: Scale refinement and validation
approach service quality assessment from these two perspectives to
develop a scale to measure robotic service quality. The following section 3.2.1. Sample and data collection
describes the scale development procedure. To refine these items, an online survey was conducted with Austra­
lian consumers who had experienced robotic service. The Qualtrics
3. Method platform was used for data collection. The survey was designed to ensure
that respondents were unable to skip questions. The target respondents
Both qualitative and quantitative methods were employed to deter­ must have used a robotic service within the last 12 months. To ensure
mine the items and dimensionality of robotic service. In accordance with the appropriateness of the sample, screening questions were developed
conventional scale development protocols, three stages of development to help prospective respondents understand the research purpose and
were undertaken across three countries (Australia, Vietnam, and China): filter those who did not fulfil the study criteria. Virtual snowball sam­
1) item generation, 2) scale refinement and validation, and 3) cross- pling was utilised for this study. This method relies on the virtual and
validation. Details of three stages are now discussed. social networks of participants and has the advantage of accessing hid­
den or hard-to-reach populations, hence, likely increasing the sample
3.1. Stage 1: Item generation and initial purification size and representativeness (Baltar and Brunet, 2012). Incentives (e.g.
gift vouchers) were provided to prospective respondents to encourage
3.1.1. Domain specification participation (an anonymous link generated by Qualtrics) through their
Prior to identifying the dimensions and items for the assessment of social media networks (e.g. Facebook). A total of 380 useable ques­
robotic service quality, the domain of robotic service was specified tionnaires were obtained. Of the respondents, 49.2% were male, 48.2%
(Churchill, 1979). First, relevant literature (including blogs, magazines, were between the ages of 26–45, and 59.7% held a bachelor’s degree or
news, and industry reports) on artificial intelligence and robots were higher. Table 1 presents the demographic information of the
reviewed. Second, service robots were categorised according to physical respondents.
and virtual forms. The former refers to frontline robots, including hu­
manoid and non-humanoid robots, and can be symbolised as frontline
service employees. The latter refers to virtual robots, symbolising em­ 3.3. Findings
ployees offering remote or online services.
After reviewing existing service quality measures, SERVQUAL (Par­ Following the suggestion of Hair et al. (2014), Schumacker and
asuraman et al., 1991) and e-SERVQUAL (Zeithaml et al., 2002) were Lomax (2004), and Costello and Osborne (2005), the dataset was
drawn upon to determine the domains of robotic service quality. The randomly split into two subsets, one for exploratory factor analysis
SERVQUAL model has five core dimensions: reliability, assurance, (EFA) and the other for confirmatory factor analysis (CFA) (see Table 2).
empathy, responsiveness, and tangibility. The first four dimensions The remaining items were factor analysed through promax oblique
correspond to employee service with an emphasis on service prompt­ rotation in SPSS 26. This analysis revealed four factors after removing 6
ness, accuracy, consistency, and employee friendliness and caring. The items with low factor or cross loadings (see Appendix for original factor
last dimension relates to the physical setting of the service, including the loadings of the eighteen items tested). The four-factor solution
employees’ appearance and other tangible components. This model was accounted for 83.57 of variation. The Kaiser-Mayer-Olkin value was
drawn upon to understand the encounter between robots and customers. 0.90, revealing satisfactory sampling adequacy. The primary loadings of
Zeithaml et al.‘s e-SERVQUAL has four core dimensions (efficiency, all the remaining items were above 0.5 and the secondary loadings were
reliability, fulfilment, and privacy) and was adapted to determine vir­ less than 0.3. The coefficient alpha for each factor was above 0.7,
tual robotic service. As robots provide information-based services, indicating adequate internal consistency (Nunnally, 1994).
Wixom and Todd’s (2005) integrated user satisfaction and technology Based on the contents of the remaining items, four factors were
acceptance scale was adapted to represent the informational and tech­ labelled as automisation, personalisation, precision, and efficiency.
nological aspects of robotic service. After scrutinising these scales and Automisation had the highest eigenvalue (7.21), explaining 57.30% of
perusing web information relating to robots and artificial intelligence, the variance in robotic service quality, whereas precision has the lowest
three domains were determined to represent robotic service quality: eigenvalue (1.06) and 5.23% of the variance. CFA was performed in
automated service, humanised service, and informational service. AMOS 26 to confirm dimensionality and assess validity. The average
Automated service represents mechanical robotic service such as mobile variance extracted for all factors were above 0.50, indicating adequate
check-in/out in the hotel. Humanised service represents intuitive and
empathetic services delivered by humanoid and non-humanoid robots, Table 1
for instance, concierge robots, robot bartenders, robot chefs, porter ro­ Demographic profile (N = 380).
bots, and vacuum cleaning robots. Informational service represents Characteristics Australia Vietnam China
analytical robots that provide accurate and up to date information (such Gender Male 187 (49.2) 219 (63%) 233 (55%)
as chatbots). This process generated 54 items that represent robotic Female 193 (50.8) 129 (37%) 190 (45%)
service. Age 18–35 142 (37.4) 192 (55%) 332 (79%)
35–46 183 (48.2) 139 (40%) 87 (20%)
55 (14.4) 19 (5%) 4 (1%)
3.1.2. Item generation
>55
Education < High school 5 (1.3) 10 (3%) 17 (4%)
Two artificial intelligence experts and three operation managers Vocational school 66 (17.4) 53 (15%) 61 (14%)
from three hotels that have used robotic service were consulted to Bachelor 76 (20) 105 (30%) 300 (71%)
ensure the appropriateness and relevance of the items. As a result, 32 Master or higher 141 (37.1) 155 (44%) 43 (10%)
Others 86 (24.2) 27 (8%) 2 (1%)
items remained. As service quality is formed on customers’ perceptions
and evaluations, 30 hotel customers who had used robotic services were Note: Values in the brackets are percentage.

3
C. Prentice and M. Nguyen Journal of Retailing and Consumer Services 62 (2021) 102661

Table 2 criterion (AIC) (Hu and Bentler, 1995), Model 3 was the most accurate
Item descriptions and measurement model results for robotic service quality. and parsimonious with the lowest AIC value (see Table 3).
Scale items Pattern Variance Cronbach’s CR AVE The nomological validity of robotic service quality was examined by
coefficient explained alpha assessing the inter-factor correlations among the four factors. The results
Automation β = 7.21 57.30% .86 .85 .65 show significant correlations ranging from 0.51 to 0.76 (see Table 4).
Robots operates .77 The criterion validity was tested by examining the relationship between
reliably AI service quality as the second-order construct and customer satisfac­
Robots performs .85 tion, since the two are related as shown in the literature (Prentice, 2013;
effectively
Robots functions .79
2014, Prentice, 2019). A customer satisfaction measure was adopted
dependably from Wixom and Todd (2005). The standardised path coefficient be­
CFI = 1.00; TLI = tween robotic service quality and customer satisfaction was 0.69 (p <
1.00; IFI = 1.00; .001).
SRMR = .00
Personalisation β = 1.18 6.13% .89 .88 .72
Robotic service is .85 3.4. Internal cross-validation
adaptive to meet a
variety of my needs
The analyses show content validity, convergent validity, discrimi­
Robotic service is .87
flexibly adjusted to nant validity, nomological validity, and criterion validity (Hair et al.,
meet my new 2014). To cross-validate the scale, full metric and scalar invariance was
demands assessed. The factor loadings were freely estimated for the two subsets of
Robotic service is .70 data in Model 1 and then constrained to be equal between the two
versatile in
addressing my
groups in Model 2. The results in Table 5 show that the measurement
needs model was invariant (Δ χ2 (11) = 12.03, p = .36; ΔCFI=.00), thus, full
CFI = 1.00; TLI = metric invariance was established.
1.00; IFI = 1.00; For full scalar invariance, the dataset was divided into two groups by
SRMR = .00
gender (see Wong and Fong, 2012). The full metric invariance by gender
Efficiency β = 2.34 11.08% .91 .91 .77
Robots are very .90 was examined. The results in Model 3 and 4 in Table 5 reveal that the
responsive to my factor loadings of the two groups were invariant (Δ χ2 (11) = 18.92, p =
requests .06; ΔCFI=.00). A latent mean analysis was performed by constraining
Robots provide .82 the intercepts of the structural equation of the observed variables on the
service in a timely
manner
latent factors which were equivalent across the two groups. The results
Robots solve my .91 in Model 5 suggest that the model fit of the latent mean model was not
problems effectively significantly different from the full metric invariant model (Δ χ2 (10) =
CFI = 1.00; TLI = 12.50, p = .25; ΔCFI=.00), thus, full scalar invariance was achieved.
1.00; IFI = 1.00;
Results of the two invariances tests confirmed a strong factorial invari­
SRMR = .00
Precision β = 1.06 5.23% .85 .83 .63 ance of the robotic service quality scale.
Information from .83
robotic system is
3.5. External cross-validation
accurate
Information from .85
robotic system is The literature contends that a new scale should be cross validated to
reliable be replicable and generalisable. Additional data was collected from
Information from .68
Vietnam (350 useable responses) and China (423 useable responses)
robotic system is
always up to date
employing a virtual snowball sampling method. Table 1 shows the de­
CFI = 1.00; TLI = mographics of the two datasets.
1.00; IFI = 1.00; Confirmatory factor analysis was performed to assess the measure­
SRMR = .00 ment model for robotic service quality. The results show good model fit
β = eigenvalues; CR = composite reliability; AVE = average variance extracted. indices for both countries (Vietnam: χ2/df = 1.80, CFI = .97, RMSEA =

convergence (Fornell and Larcker, 1981). The composite reliability for Table 4
each factor was higher than the cut-off level of 0.70 as shown in Table 2. Correlation coefficients, means, and standard deviations of robotic service
As suggested by Brakus et al. (2009), three competing models were quality and customer satisfaction.
analysed to examine the factor structure for this scale: 1) the one-factor Means SD 1 2 3 4
model with all items loading on a single factor; 2) the first-order factor 1.Automisation 4.93 1.10
with the four dimensions as independent factors; and 3) the 2.Personalisation 4.94 1.12 .72**
second-order factor with robotic service quality as the higher order. The 3.Efficiency 4.10 1.53 .59** .51**
results in Table 3 show that the second-order factor structure had the 4. Precision 4.95 1.12 .76** .69** .57**
5.Cust. satisfaction 4.81 1.27 .66** .60** .61** .60**
best model fit indices: χ2 = 86.19, def = 46, χ2/df = 1.87, CFI = .99,
RMSEA = .05, SRMR = .02. Consistent with the Akaike’s information **p < .01.

Table 3
Confirmatory factor analysis model fit comparisons.
Model Description χ2 df χ2/df p CFI RMSEA SRMR AIC

1 One factor 550.77 54 10.20 .00 .86 .16 .05 598.77

2 First-order factor 125.79 38 3.31 .00 .98 .08 .03 181.79
3 Second-order factor 86.19 46 1.87 .00 .99 .05 .02 150.19

4
C. Prentice and M. Nguyen Journal of Retailing and Consumer Services 62 (2021) 102661

Table 5
Test of measurement model invariance.
Model Description Compare with χ2 Df Δ χ2 (Δdf) p CFI ΔCFI RMSEA SRMR

1 Baseline model (G1/2) – 154.20 90 – – .98 – .05 .03

2 Factor pattern (G1/2) Model 1 169.66 101 12.03(11) .36 .98 .00 .04 .04
3 Baseline model (M/F) – 157.19 90 – – .98 – .04 .03
4 Factor pattern (M/F) Model 3 176.11 101 18.92(11) .06 .98 .00 .04 .03
5 Scalar invariance (M/F) Model 4 188.61 111 12.50(10) .25 .98 .00 .04 .04
6 Based model (Aus/VN) – 300.50 90 – – .96 – .06 .04
7 Factor pattern (Aus/VN) Model 6 314.05 101 13.55(11) .26 .96 .00 .04 .04
8 Based model (Aus/CN) – 240.12 90 – – .97 – .05 .04
9 Factor pattern (Aus/CN) Model 8 255.83 101 15.71(11) .15 .97 .00 .04 .05

Note: M = male, F = female, Aus = Australia, VN- Vietnam, CN-China.

.07, SRMR = .03; China: χ2/df = 3.39, CFI = .95, RMSEA = .08, SRMR = literature to reflect customers’ perception and evaluation of robotic
.04). To assess full metric invariance, datasets from Australia and Viet­ service quality. Through a series of scale development processes, four
nam, and from Australia and China were used to perform a group dimensions were identified to represent robotic service quality:
invariance test by constraining the factor-loading estimates to be equal automisation, personalisation, precision, and efficiency. These di­
between the two groups. The results in Model 6 and 7 in Table 5 reveal mensions are indicative of mechanical, analytical, intuitive, and empa­
that the model was invariant across the different sample cohorts be­ thetic artificial intelligence proposed by Huang and Rust (2018).
tween Australia and Vietnam (Δ χ2 (11) = 13.55, p = .26; ΔCFI=.00). Automisation refers to the level of automation and performance by
The results in Model 8 and 9 reveal that the model was invariant be­ robots. This dimension mainly captures customers’ experience with AI
tween the Australian and Chinese data (Δ χ2 (11) = 15.71, p = .15; system quality. Sluggish operating systems underlying the robotic ser­
ΔCFI=.00). These results support the replicability of the scale. vices evokes a negative customer response. Advanced technology today
Cross validation among the three datasets was conducted. Re­ has changed customers’ expectations of technology-related services.
spondents (350) were randomly selected from data collected in China Personalisation indicates how robots can be flexible, like service em­
and Australia to validate the robotic service quality scale across coun­ ployees, to meet customer’s requests and demands. Service robots are
tries. The relationship between robotic service quality as the second- designed to represent and behave like humans. Customers expect their
order construct and user satisfaction were tested across the three data­ interactions with these robots to be similar to those with service em­
sets, Australia, Vietnam, and China (see Table 6). The standardised path ployees. Robots represent these employees and provide the requested
coefficients between robotic service quality and customer satisfaction information and function like employees. Customers may have different
were significant across the three datasets, supporting the external val­ needs and wants from the service provider. Whilst employees’ adapt­
idity of robotic service quality measurement. ability is expected, they are nevertheless limited to fixed working hours
and oft times, inadequate training. Human physical and emotional
4. Discussion constraints do not affect machine operated robots. These robots are
available 24/7 and are capable of storing infinite amounts of informa­
This study draws upon service quality conceptualisation and devel­ tion that can be adapted to meet customers’ requests.
oped a scale to measure robotic service quality. Service quality is formed Precision is reflective of the accuracy and exactness of the informa­
as a result of customers’ service encounter experience with the service tion provided by robots. Such information is programmed through big
provider. The service encounter, within the service quality literature, data analytics and machine learning without unintended human inter­
primarily refers to customers’ interaction with service employees and vention. Information generated by robots must be achieved in a timely
the online or physical settings associated with the service organisation. manner and outdated information can be misleading. Customers expect
Existing service quality measures fail to capture the services provided by information from robotic system to be updated, objective, thorough, and
robots that have been widely used in service industries. Robots may have accurate. Precise information can guide customers to make informed
physical and virtual forms. The former includes humanoid robots that decisions on consumption and purchase activities. Efficiency indicates
look and behave like humans (e.g. Connie) and non-humanoid robots (e. timely and responsive service from robots. The ability to respond to
g. mobile robots). The latter refers to online automated services. Service customer requests dependably and reliably affects customer experience,
organisations utilise robotic services to improve customer experience attitudes, and subsequent behaviours.
and generate positive customer response. This study views robotic ser­
vices as a service product and developed a robotic service quality scale. 5. Implications
The interaction between customers and humanoid robots is referred to
as a personal encounter, interactions between customers and other 5.1. Theoretical implications
forms of robots are impersonal encounters.
Given the services provided by robots is largely information- and This study developed a scale to measure robotic service quality.
technology-based, the items that were generated to develop the scale Establishment of this scale enriches the service quality and marketing
were drawn from the service quality and technology acceptance literature. Numerous scales have been developed to measure service
quality. Some are modelled to measure generic service quality (e.g.
SERVQUAL, SERVPERF), others are more context based, for instance, e-
Table 6 travel service quality (Ho and Lee, 2007), retail service quality (Dab­
External cross-validation. holkar et al., 1996), bank service quality (Karatepe et al., 2005), e-ser­
Customer satisfaction vice quality (Li and Suomi, 2009), web information service quality
Australia Vietnam China
(Yang et al., 2005), and healthcare service quality (Dagger et al., 2007).
Given the pervasive use of robots for customer service in various service
Robotic service quality Australia .78*** .31* .85***
industries, it is imperative to understand how robotic service quality
Vietnam .50*** .57*** .44***
China .70*** .15* .68*** may affect customer attitudes and behaviours.
Service robots can deliver service similarly or beyond that of service
*p < .05, **p < .01, ***p < .001.

5
C. Prentice and M. Nguyen Journal of Retailing and Consumer Services 62 (2021) 102661

employees. Therefore, robotic service quality can complement the other

service quality scales in industries where robots are used. For instance,
robots in HSBC can greet customers, guide them on the selection of
appropriate banking services and summon an employee to help with
transactions. The robot “Pepper” (Fig. 1) can also pose for selfies and tell
jokes. Robotic service quality can be used in conjunction with Karetepe
et al.’s (2005) service quality of retail banks to measure bank service
quality and understand its impact on customer satisfaction and
behaviour.
Robotic service quality can also be integrated with Parasuraman
et al.‘s SERVQUAL, which consists of five dimensions (tangibility, reli­ Fig. 1. SoftBank Robotics’ humanoid robot, Pepper in HSBC at the Fifth Ave
branch on Monday, June 25, 2018 in New York. Mark Von Holden | AP Images
ability, assurance, responsiveness, and empathy). Each has relevance to
for HSBC.
service employees (Prentice, 2013). This scale has been widely used in
the hospitality industry (Gabbie and O’Neill, 1996; Raspor, 2010; Saleh
and Ryan, 1991; Shi et al., 2014). Robots in hotels (see Fig. 2) are able to
interact with guests and provide the requested information. In addition
to measuring customer perceptions of hotel service quality on the five
SERVQUAL dimensions, robotic service quality can be assessed to gain a
wholistic picture of hotel service quality and its influence on customer
experience (see Fig. 2).
Given that some robotic services are online information-based ser­
vices, robotic service quality can also be integrated to e-service quality
to present a broader perspective of the quality of web-based services.
Existing e-service quality scales (Dabholkar et al., 1996; Madu and Fig. 2. Connie in Hilton interacting with hotel guests.
Madu, 2002; Zeithaml et al., 2002) fail to capture the service provided
by chatbots that are available to answer queries and provide information 6. Limitations
around the clock. Chatbot services are ubiquitous and the topic is
becoming increasingly popular in the literature (Dale, 2016; Feine et al., Despite endeavours to the contrary, limitations within this study are
2019; Jones and Jones, 2019; Lasek and Jessa, 2013). Understanding the acknowledged. First, the sample frame was not representative of every
service quality of chatbot services is conducive to improving customer service industry. Thus, the generalisability of this scale is limited. The
experience. Robotic service quality addresses this void in the chatbot sample frame could be extended to better represent the targeted popu­
literature. lation. Second, robotic service varies across different service contexts
The study also adds a new measure to services and relationship and industries. Development of this scale should differentiate between
marketing to manage customer relationships. Traditional service mar­ services to gain greater insights. Third, cross-validation in different
keting strategies are primarily focused on the 7Ps (product, promotion, countries should have taken the cultural effect into account. Finally, this
price, place, people, process, and physical evidence) and manifested in study did not include a B2B context in the scale development process. In
defensive (or relationship marketing) and aggressive approaches practice, robots have been used in these contexts to improve overall
(Prentice, 2019). Service quality is considered a relationship marketing service quality for business customers. Future research should address
technique to elicit customer satisfaction and loyalty. Robotic service these limitations.
quality can be incorporated into the services and relationship marketing
literature to better understand customer attitudes and behaviours. References

5.2. Practical implications Akter, S., D’Ambra, J., Ray, P., 2013. Development and validation of an instrument to
measure user perceived service quality of mHealth. Inf. Manag. 50 (4), 181–195.
https://doi.org/10.1016/j.im.2013.03.001.
Robotic service quality as developed in this study can be applied to Baltar, F., Brunet, I., 2012. Social research 2.0: virtual snowball sampling method using
relevant industries and organisations that extensively use service robots Facebook. Internet Res. 22 (1), 57–74. https://doi.org/10.1108/
10662241211199960.
to provide customer service. Robots are mostly designed and developed Bartneck, C., Kanda, T., Mubin, O., Al Mahmud, A., 2009. Does the design of a robot
by IT experts. Most users may be layman and not understand the un­ influence its animacy and perceived intelligence? International Journal of Social
derlying principles and operating systems. This scale helps service pro­ Robotics 1 (2), 195–204.
Belpaeme, T., Kennedy, J., Ramachandran, A., Scassellati, B., Tanaka, F., 2018. Social
viders and customers appreciate the importance of robotic service robots for education: a review. Science robotics 3 (21), 1–9.
without understanding the complex IT construction. Service providers Belanche, D., Casaló, L.V., Flavián, C., 2020. Frontline robots in tourism and hospitality:
can use this scale to assess return on investment as the purchase of robots service enhancement or cost reduction? Electron. Mark. 1–16.
Boshoff, C., 2007. A psychometric assessment of E-S-QUAL: a scale to measure electronic
can be expensive. Service providers may also gain a better understand­
service quality. J. Electron. Commer. Res. 8 (1), 101.
ing of the quality of these tools and their impact on customer service. Brady, M.K., Cronin, J.J., 2001. Some new thoughts on conceptualizing perceived service
Robotic service quality can help providers address the advantages and quality: a hierarchical approach. J. Market. 65 (3), 34–49. https://doi.org/10.1509/
jmkg.65.3.34.18334.
disadvantages of robotic service. From customer perspective, robotic
Brakus, J.J., Schmitt, B.H., Zarantonello, L., 2009. Brand experience: what is it? How is it
service quality can be a useful tool for them to communicate with service measured? Does it affect loyalty? J. Market. 73 (3), 52–68. https://doi.org/10.1509/
providers to address their appreciation and concerns. This scale can also jmkg.73.3.52.
assist customers to provide feedback to the service provider. For Brochado, A., 2009. Comparing alternative instruments to measure service quality in
higher education. Qual. Assur. Educ. 17 (2), 174–190. https://doi.org/10.1108/
example, some information provided by chatbots can be limited and 09684880910951381.
unhelpful. Without proper measures, customers are unable to raise the Chuah, S.H.W., Yu, J., 2021. The future of service: The power of emotion in human-robot
issue with the service provider. This scale serves this purpose so that interaction. J. Retail. Consum. Serv. 61, 102551.
Churchill, G.A., 1979. A Paradigm for developing better measures of marketing
providers can improve robotic service quality to enhance customer constructs. J. Market. Res. 16 (1), 64–73. https://doi.org/10.2307/3150876.
experience and positive experiences. Cobos, L.M., Mejia, C., Ozturk, A.B., Wang, Y., 2016. A technology adoption and
implementation process in an independent hotel chain. Int. J. Hospit. Manag. 57,
93–105. https://doi.org/10.1016/j.ijhm.2016.06.005.

6
C. Prentice and M. Nguyen Journal of Retailing and Consumer Services 62 (2021) 102661

Collier, J.E., Bienstock, C.C., 2006. Measuring service quality in e-retailing. J. Serv. Res. Madu, C.N., Madu, A.A., 2002. Dimensions of E-quality. International Journal of Quality
8 (3), 260–275. & Reliability Management.
Costello, A.B., & Osborne, J. (2005). Best practices in exploratory factor analysis: Four Marinova, D., de Ruyter, K., Huang, M.-H., Meuter, M.L., Challagalla, G., 2017. Getting
recommendations for getting the most from your analysis. Pract. Assess. Res. smart: learning from technology-Empowered frontline interactions. J. Serv. Res. 20
Evaluat., 10(1), 7. 1-9. (1), 29–42. https://doi.org/10.1177/1094670516679273.
Cronin, J.J., Taylor, S.A., 1992. Measuring service quality: a reexamination and McDougall, G.H.G., Levesque, T.J., 1995. A revised view of service quality dimensions:
extension. J. Market. 56 (3), 55–68. https://doi.org/10.1177/ an empirical investigation. J. Prof. Serv. Market. 11 (1), 189–210. https://doi.org/
002224299205600304. 10.1300/J090v11n01_13.
Dabholkar, P.A., Thorpe, D.I., Rentz, J.O., 1996. A measure of service quality for retail Nunnally, J.C., 1994. Psychometric Theory, third ed. third ed. McGraw-Hill, New York.
stores: scale development and validation. J. Acad. Market. Sci. 24 (1), 3–16. https:// Ozment, J., Morash, E.A., 1994. The augmented service offering for perceived and actual
doi.org/10.1007/BF02893933. service quality. J. Acad. Market. Sci. 22 (4), 352–363. https://doi.org/10.1177/
Dagger, T.S., Sweeney, J.C., Johnson, L.W., 2007. A hierarchical model of health service 0092070394224004.
quality: scale development and investigation of an integrated model. J. Serv. Res. 10 Parasuraman, A., Berry, L.L., Zeithaml, V.A., 1991. Refinement and reassessment of the
(2), 123–142. https://doi.org/10.1177/1094670507309594. SERVQUAL scale. J. Retailing 67 (4), 420–450.
Dale, R., 2016. The return of the chatbots. Nat. Lang. Eng. 22 (5), 811–817. https://doi. Parasuraman, A., Zeithaml, V.A., Berry, L.L., 1985. A conceptual model of service quality
org/10.1017/S1351324916000243. and its implications for future research. J. Market. 49 (4), 41–50. https://doi.org/
Ding, D.X., Hu, P.J.-H., Sheng, O.R.L., 2011. e-SELFQUAL: a scale for measuring online 10.1177/002224298504900403.
self-service quality. J. Bus. Res. 64 (5), 508–515. https://doi.org/10.1016/j. Parasuraman, A., Zeithaml, V.A., Berry, L.L., 1988. Servqual: a multiple-item scale for
jbusres.2010.04.007. measuring consumer perc. J. Retailing 64 (1), 12.
Feine, J., Morana, S., Gnewuch, U., 2019. Measuring service encounter satisfaction with Parasuraman, A., Zeithaml, V.A., Berry, L.L., 1994. Alternative scales for measuring
customer service chatbots using sentiment analysis. In: Proceedings of the 14th service quality: a comparative assessment based on psychometric and diagnostic
International Conference on Wirtschaftsinformatik (WI2019). Paper presented at the criteria. J. Retailing 70 (3), 201–230.
in, Siegen, Germany. Prentice, C., 2013. Service quality perceptions and customer loyalty in casinos. Int. J.
Fornell, C., Larcker, D.F., 1981. Evaluating structural equation models with unobservable Contemp. Hospital. Manag. 25 (1), 49–64.
variables and measurement error. J. Market. Res. 18 (1), 39–50. https://doi.org/ Prentice, C., 2019. Emotional Intelligence and Marketing. World Scientific.
10.2307/3151312. Prentice, C., Dominique Lopes, S., Wang, X., 2020. The impact of artificial intelligence
Gabbie, O., O’Neill, M.A., 1996. SERVQUAL and the Northern Ireland hotel sector: a and employee service quality on customer satisfaction and loyalty. J. Hospit. Market.
comparative analysis - part 1. Manag. Serv. Qual.: Int. J. 6 (6), 25–32. https://doi. Manag. 29 (7), 739–756.
org/10.1108/09604529610149194. Raspor, S., 2010. Measuring perceived service quality using SERVQUAL: a case study of
Gartner, 2019. Gartner Survey Shows 37 Percent of Organizations Have Implemented AI the Croatian hotel industry. Management 5 (3).
in Some Form. Retrieved from. https://www.gartner.com/en/newsroom/press-rele Rajesh, M., 2015. Inside Japan’s First Robot-Staffed Hotel. Retrieved from. https://www.
ases/2019-01-21-gartner-survey-shows-37-percent-of-organizations-have. theguardian.com/travel/2015/aug/14/japan-henn-na-hotel-staffed-by-robots.
Gronroos, C., 1982. Strategic Management and Marketing in the Service Sector. Rouhiainen, L., 2019. How AI and Data Could Personalize Higher Education. Retrieved
Chartwell-Bratt, London. from. https://hbr.org/2019/10/how-ai-and-data-could-personalize-higher-educati
Grönroos, C., 1994. From scientific management to service management: a management on.
perspective for the age of service competition. Int. J. Serv. Indust. Manag. 5 (1), Russell, S.J., Norvig, P., ProQuest, 2016. Artificial Intelligence: a Modern Approach (Global,
5–20. 3rd; ed. Pearson Education Limited, Harlow, Essex, England.
Gursoy, D., Chi, O.H., Lu, L., Nunkoo, R., 2019. Consumers acceptance of artificially Saleh, F., Ryan, C., 1991. Analysing service quality in the hospitality Industry using the
intelligent (AI) device use in service delivery. Int. J. Inf. Manag. 49, 157–169. SERVQUAL model. Serv. Ind. J. 11 (3), 324–345. https://doi.org/10.1080/
https://doi.org/10.1016/j.ijinfomgt.2019.03.008. 02642069100000049.
Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E., 2014. Multivariate Data Analysis, Santos, J., 2003. E-service quality: a model of virtual service quality dimensions. Manag.
Seventh, Pearson new international ed. Pearson Education Limited, Harlow, Essex. Serv. Qual.: Int. J.
Helgesen, Ø., 2006. Are loyal customers profitable? Customer satisfaction, customer Schumacker, R.E., Lomax, R.G., 2004. A Beginner’s Guide to Structural Equation
(action) loyalty and customer profitability at the individual level. J. Market. Manag. Modeling, second ed. Lawrence Erlbaum Associates, Mahwah, N.J.
22 (3–4), 245–266. https://doi.org/10.1362/026725706776861226. Shi, Y., Prentice, C., He, W., 2014. Linking service quality, customer satisfaction and
Heskett, J.L., Jones, T.O., Loveman, G.W., Sasser, W.E., Schlesinger, L.A., 1994. Puting loyalty in casinos, does membership matter? Int. J. Hospit. Manag. 40, 81–91.
the service-profit chain to work. Retrieved from. https://hbr.org/2008/07/putting https://doi.org/10.1016/j.ijhm.2014.03.013.
-the-service-profit-chain-to-work, 2, vol. 72, 164-174. Siu, N.Y.M., Tak-Hing Cheung, J., 2001. A measure of retail service quality. Market.
Ho, C.-I., Lee, Y.-L., 2007. The development of an e-travel service quality scale. Tourism Intell. Plann. 19 (2), 88–96. https://doi.org/10.1108/02634500110385327.
Manag. 28 (6), 1434–1449. https://doi.org/10.1016/j.tourman.2006.12.002. Tavakoli, R., Mura, P., 2018. Netnography in tourism – beyond web 2.0. Ann. Tourism
Hu, L.-T., Bentler, P.M., 1995. Evaluating model fit. In: Hoyle, R.H. (Ed.), Structural Res. 73, 190–192. https://doi.org/10.1016/j.annals.2018.06.002.
Equation Modeling: Concepts, Issues, and Applications. Sage, Thousand Oaks, CA. West, A., Clifford, J., Atkinson, D., 2018. Alexa, Build Me a Brand" an Investigation into
Huang, M.-H., Rust, R.T., 2018. Artificial intelligence in service. J. Serv. Res. 21 (2), the Impact of Artificial Intelligence on Branding, 2018, London.
155–172. Wirtz, J., Lovelock, C., 2018. Developing Service Products and Brands. WS Professional.
Ivanov, S., Webster, C., Berezina, K., 2018. Adoption of robots and service automation by Wixom, B.H., Todd, P.A., 2005. A theoretical integration of user satisfaction and
tourism and hospitality companies. Revista Turismo & Desenvolvimento 27/28, technology acceptance. Inf. Syst. Res. 16 (1), 85–102. https://doi.org/10.1287/
1501–1517. isre.1050.0042.
Jones, B., Jones, R., 2019. Public service chatbots: automating conversation with BBC Wong, I.A., Fong, V.H.I., 2012. Development and validation of the casino service quality
news. Digital Journalism 7 (8), 1032–1053. https://doi.org/10.1080/ scale: CASERV. Int. J. Hospit. Manag. 31 (1), 209–217. https://doi.org/10.1016/j.
21670811.2019.1609371. ijhm.2011.04.005.
Kaplan, A., Haenlein, M., 2019. Siri, Siri, in my hand: who’s the fairest in the land? On Woodside, A.G., Frey, L.L., Daly, R.T., 1989. Linking sort/ice anlity, customer
the interpretations, illustrations, and implications of artificial intelligence. Bus. satisfaction, and behavioral intention. J. Health Care Market. 9 (4), 5–17.
Horiz. 62 (1), 15–25. https://doi.org/10.1016/j.bushor.2018.08.004. Wu, T., Tegmark, M., 2019. Toward an artificial intelligence physicist for unsupervised
Karatepe, O.M., Yavas, U., Babakus, E., 2005. Measuring service quality of banks: scale learning. Phys. Rev. E 100 (3), 1–19, 033311.
development and validation. J. Retailing Consum. Serv. 12 (5), 373–383. https:// Xu, G., Mu, Y., Liu, J., 2017. Inclusion of artificial intelligence in communication
doi.org/10.1016/j.jretconser.2005.01.001. networks and services. ITU J., ICT Discover. Spec. 1, 1–6.
Kim, H.J., 2011. Service orientation, service quality, customer satisfaction, and customer Yang, Z., Cai, S., Zhou, Z., Zhou, N., 2005. Development and validation of an instrument
loyalty: testing a structural model. J. Hospit. Market. Manag. 20 (6), 619–637. to measure user perceived service quality of information presenting Web portals. Inf.
https://doi.org/10.1080/19368623.2011.577698. Manag. 42 (4), 575–589. https://doi.org/10.1016/S0378-7206(04)00073-4.
Kumar, K.N., Balaramachandran, P.R., 2018. Robotic process automation - a study of the Yang, Z., Jun, M., Peterson, R.T., 2004. Measuring customer perceived online service
impact on customer experience in retail banking industry. J. Internet Bank. Commer. quality. Int. J. Oper. Prod. Manag. 24 (11), 1149–1174. https://doi.org/10.1108/
23 (3), 1–27. 01443570410563278.
Lasek, M., Jessa, S., 2013. Chatbots for customer service on hotels’ websites. Inf. Syst. Zeithaml, V.A., 2000. Service quality, profitability, and the economic worth of
Manag. 2 (2), 146–158. customers: what we know and what we need to learn. J. Acad. Market. Sci. 28 (1),
Li, H., Suomi, R., 2009. A proposed scale for measuring e-service quality. International 67–85. https://doi.org/10.1177/0092070300281007.
Journal of u-and e-Service, Science and Technology 2 (1), 1–10. Zeithaml, V.A., Parasuraman, A., Malhotra, A., 2002. Service quality delivery through
Lee, G.G., Lin, H.F., 2005. Customer Perceptions of E-service Quality in Online Shopping. web sites: a critical review of extant knowledge. J. Acad. Market. Sci. 30 (4),
International Journal of Retail & Distribution Management. 362–375. https://doi.org/10.1177/009207002236911.