Modeling customer satisfaction and revisit intention from online restaurant reviews: an attribute-level analysis
DOI | https://doi.org/10.1108/IMDS-09-2022-0570 |
Published date | 28 March 2023 |
Date | 28 March 2023 |
Pages | 1548-1568 |
Author | Futao Zhao,Hao Liu |
Modeling customer satisfaction
and revisit intention from
online restaurant reviews:
an attribute-level analysis
Futao Zhao and Hao Liu
Northeastern University at Qinhuangdao, Qinhuangdao, China
Abstract
Purpose –The purpose of thispaper is to detect predefined service attributes and their sentiments from online
restaurant reviews, and then to measure the effects of customer sentiments toward service attributes on
customer satisfaction (CS) and revisit intention (RVI) simultaneously.
Design/methodology/approach –This study proposed a supervised framework to model CS and RVI
simultaneously from restaurant reviews. Specifically, the authors detected the predefined service dimensions
from online reviews based on random forest. Then, the sentiment polarities of the reviews toward each
predefined dimension were identified using light-gradient boosting machine (LightGBM). Finally, the effects of
attribute-specific sentiments on CS and RVI were evaluated by a bagged neural network-based model. The
proposed framework was evaluated by 305,000 restaurant comments collected from DianPing.com, a Yelp-like
website in China.
Findings –The authors obtained a hierarchal importance order of the investigated service themes (i.e.
location, service, environment, price and food). The authors found that food played the most important role in
affecting both CS and RVI. The most salient attribute with respect to each service theme was also identified.
Originality/value–Unlike prior work relying on the data collected from surveys, this study is among the first
to model the relationship among service attributes, CS and RVI simultaneously from real-world data. The
authors established a hierarchal structure of eighteen attributes within five service themes and estimatedtheir
effects on both CS and RVI, which will broaden our understanding of customer perception and behavioral
intention during service consumption.
Keywords Online reviews, Text mining, Service attributes, Customer satisfaction, Revisit intention
Paper type Research paper
1. Introduction
In the context of hospitality and service, unveiling the impacts of service quality attributes on
customer satisfaction (CS) and revisit intention (RVI) plays an important role in developing
management insight and improving service delivery, especially in the restaurant sector (Mejia
et al., 2021). As such, it has drawn aggressive attention from researchers in this field. However,
there exist challenges in establishing a clear relationship among these variables about consumer
behavior and restaurant performance (Vencovsk
y, 2020). In comparison with product quality,
service quality is hard to evaluate due to its intrinsic nature suchas intangibilityand variability
(Christian, 1984). To tackle this challenge, two survey approaches, including questionnaires and
interviews, are usually applied to measure service quality from the viewpoint of customer
experience (Berry et al., 1998;Parasuramanet al., 1985). Although the measurement scales used
in the surveys such as DINESERV are well-established, the approaches have twofold
shortcomings accompanied by their prevalent applications. First of all, it is challenging to
IMDS
123,5
1548
This work has been supported by the National Natural Science Foundation of China (Grant no.
72274032), the Fundamental Research Funds for the Central Universities (Grant no. N2223033), the
Humanities and Social Science Research Project of Hebei Education Department (Grant no. BJ2021104),
the Natural Science Foundation of Hebei Province (Grant no. G2021501012), and the Social Science
Foundation of Liaoning Province (Grant no. L20BXW004).
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/0263-5577.htm
Received 26 September 2022
Revised 14 January 2023
Accepted 3 March 2023
Industrial Management & Data
Systems
Vol. 123 No. 5, 2023
pp. 1548-1568
© Emerald Publishing Limited
0263-5577
DOI 10.1108/IMDS-09-2022-0570
provide generalized and scalable measures because of the heterogeneous consumer preferences
and time-varying restaurant operations (Mejia et al.,2021). Second, the costs of conducting
surveysare huge in bothtime and labor. Therefore, moreacademic attemptsneed to be made on
service evaluation via exploring other alternative channels.
With the advent of digital platforms, consumer-generated content presented on online
review websites opens up new avenues for measuring customer experience and service
quality (Chen et al., 2020;Schuckert et al., 2015). Third-party platforms like Yelp and DianPing
allow consumers to post online reviews about post-purchase feedback, which in turn
facilitates prospective consumers to make informed decisions. Prior work has highlighted the
value of online reviews in many research fields such as product design (Bi et al., 2019;Qi et al.,
2016), market segmentation (Ahani et al., 2019) and trust perception (Cheng et al., 2019).
Moreover, a substantial body of literature has attempted to mine service dimensions, CS and
RVI from service-related reviews, particularly in the hotel and airline industries (Bi et al., 2020;
Lucini et al., 2020;Luo and Tang, 2019;Park, 2019;Park et al., 2020a,2020b;Zhao et al., 2015,
2019). In contrast to surveys, online review data not only contain realistic, spontaneous and
comprehensive information regarding consumer experience (Vidal et al., 2016), but also can
provide restaurant managers with operational advantages derived from its large volume,
easy availability and low cost (Tian et al., 2021). In this regard, we argue that online reviews
can serve as an alternative potential information source to effectively understand service
attributes, customer experience and behavioral intention in the restaurant industry.
We conducteda comprehensivereview of relatedpublicationsand found thatlimited and yet
fast-expanding studies have used online review data to analyze dining experience in a more
complete and up-to-date manner (Gan et al., 2017;Jia, 2020;Kim et al., 2020;Liu et al., 2020;Mejia
et al., 2021;Nakayama and Wan, 2018;Tian et al.,2021;Vuet al., 2019;Yan et al., 2015;Zhu et al.,
2019). To the best of our knowledge, two research gaps are existing in the relatively scarce
literature. Theoretically, the majority of these studies focused on exploring the effect of
restaurant service attributes on overall CS. Despite the relevance of retaining existing customers
to the restaurant performance (Kim et al., 2009), few attempts have been made to capture RVI
from onlinerestaurantreviews. Methodologically,amid the wealthyand diversifiedinformation
of online reviews, prior research typically used only review metadata such as numerical ratings
to explore food-related sentiments (Liu et al.,2020;Yan et al., 2015). Moreover, although some
extant studies have concentrated on the textual review content, most of them adopted
techniquessuch ascontent analysisand topic modelto uncover theservice dimensionshidden in
the texts (Mejia et al., 2021;Tian et al., 2021). These used methods, which can be categorized into
unsupervised ways, might be not applicable in supervised scenarios where a set of predefined
service dimensions are given by restaurateurs or platforms.
The current research was therefore conducted to overcome the two limitations by
proposing a supervised framework to model CS and RVI simultaneously from textual
reviews. Specifically, our research was motivated by the following facts. First, prior research
usually used online customer rating scores to represent CS. However, it was difficult to find a
similar proxy for RVI. Textual reviews had the potential to represent the two variables with
accurate values. Moreover, few studies focused on CS and RVI simultaneously, and it was still
unclear about the difference between the impacts of service quality on the two important
service outcome variables, which might be valuable for the service operation strategies.
Finally, most of the previous work used unsupervised methods to mine service attributes
from reviews. Given a certain service attributed provided by platforms or restaurant
managers, previous methods failed to measure its impact on CS and were not suitable for this
situation. In sum, this study aims to fulfill the following research objectives:
RO1. To detect predefined service dimensions and their sentiments from online
restaurant reviews based on supervised learning.
Customer
satisfaction
and revisit
intention
1549
To continue reading
Request your trial