Monitoring and improving Greek banking services using Bayesian Networks: An analysis of mystery shopping data
Highlights
► We use Bayesian Networks (BNs) to analyse data gathered from mystery shoppers’ report. ► We evaluate the quality of service offered by the loan departments of Greek Banks. ► The key factors that influence global satisfaction are identified via BNs. ► Using BNs real time results regarding the effect improvement action are obtained.
Introduction
The banking industry is a highly competitive and customer oriented organisation. Customer retention and attraction is a core element of its managing strategy; customer service is one of the factors allowing to differentiate a bank from its competitors.
Roughly speaking, customer satisfaction refers to the extent to which products and services supplied by a company meet or exceed customer expectation. Customer satisfaction levels can be measured using survey techniques and evaluation questionnaires. High levels of customer satisfaction indicate a good performance of the business since satisfied customers are most likely to be loyal to the specific company and use a wide range of services. Understanding which elements influence customer satisfaction is important not only to describe the actual situation but also to plan and implement possible improvement actions.
In this paper we use Bayesian Networks (BN hereafter) to analyse data gathered from mystery shoppers’ report. To our knowledge, this is the first time that these techniques are used in combination. We present a real data analysis concerning customer evaluation of service provided by the loan unit of Greek Banks. For some recent works regarding customer satisfaction analysis of Greek Banks see e.g. Mihelis et al., 2001, Grigoroudis et al., 2002, Mylonakis, 2009, Kagara and Voyiatzis, 2010.
Mystery shopping is a well known marketing technique used by companies and marketing analysts to measure quality of service, and gather specific information about products and services. Nowadays, it is one of the most used techniques for performance evaluation of banks; see e.g. Schrader, 2006, Sherman and Zhu, 2006, Roberts and Campbell, 2007 and references therein.
A BN is a probabilistic graphical model that represents a set of random variables and their conditional dependencies via a Directed Acyclic Graph (DAG). The use of a graph, as a pictorial representation of the problem at hand, simplifies model interpretation, and facilitates communication and interaction among experts with different backgrounds. For these reasons, BNs are widely applied in different fields for the analysis of multivariate data, see Neapolitan (2004).
Recently BNs have been successfully applied to the analysis of customer satisfaction data, see for example Salini and Kenett, 2007, Renzi et al., 2009. Providing a DAG representation for the problem under investigation, BNs allow to easily identify the key elements influencing customer satisfaction. Furthermore they can be used to simulate improvement strategies, getting reliable results in a straightforward manner.
The paper is organised as follows. In Section 2.1 we present the mystery shopping methodology. BNs together with the procedure to construct them are illustrated in Sections 2.2 Bayesian Networks, 2.3 Structural learning for BNs. Section 3 is devoted to the application of BNs to service quality improvement in Greek Banks. Finally, in Section 4 we end up with some comments and final remarks.
Section snippets
Mystery shopping
Mystery shopping is a well established methodology which was introduced in the early 1940s primarily by the management of banks and retail chain stores to assess the integrity of their employees (Zikmund, Babin, Carr, & Griffin, 2009). Nowadays there are hundreds of companies providing services related to mystery shopping surveys (see for example http://www.mysteryshop.org/). For a comprehensive introduction and an exhaustive discussion on the topic we refer the reader to the publications of
Data description
The mystery shopping data were collected by a female master student of the University of Aegean (Sergianniti, 2003). The aim of the research was to evaluate and compare the quality of the services offered by loan units of five popular banks operating in Greece in 2003. The names of the banks are suppressed for privacy reasons but they are available in Sergianniti (2003). The data consist of 128 mystery shopping reports conducted from late April to late August 2003 in five different cities:
Concluding remarks
Service performance and its impact on the customer experience are key factors for bank management. Customers are free to choose between competitive alternatives, therefore companies should pay attention not only to the quality of service provided but also to its effectiveness. One method for service evaluation, that has increased in popularity in recent years, is the use of mystery shoppers. Mystery shoppers are “fake customers” used to survey and monitor the quality of the service and to
Acknowledgements
The work of the first author was partially supported by MIUR, Italy, PRIN 2007XECZ7L, and the University of Pavia. The work of the second author was partially supported by MIUR, Italy, PRIN 2007BJK2PT, and the University Roma Tre. The authors are grateful to Mrs. Anna Sergianiti for providing us with the data and conducting the mystery shopping visits.
References (22)
- et al.
Learning Bayesian networks from data: An information-theory based approach
Artificial Intelligence
(2002) The EM algorithm for graphical association models with missing data
Computational Statistics and Data Analysis
(1995)- et al.
Customer satisfaction measurement in the private bank sector
European Journal of Operational Research
(2001) - et al.
Approximating discrete probability distributions with dependence trees
IEEE Transactions on Information Theory
(1968) - et al.
A Bayesian method for the induction of probabilistic networks from data
Machine Learning
(1992) - et al.
Probabilistic networks and expert systems
(1999) - et al.
Satisfaction benchmarking and customer classification: An application to the branches of a banking organization
International Transactions in Operational Research
(2002) - Kagara, E., & Voyiatzis, C. (2010). The customer service in the banking sector as a means of creating a competitive...
An introduction to Bayesian Networks
(1996)Graphical models
(1996)
Customer relationship management functions: A survey of Greek Bank customer satisfaction perceptions
The IUP Journal of Bank Management
Cited by (22)
Network models for cyber attacks evaluation
2023, Socio-Economic Planning SciencesSocial media big data integration: A new approach based on calibration
2018, Expert Systems with ApplicationsCitation Excerpt :For examples of BN applications to study management efficiency, web site usability, operational risks, biotechnology, customer satisfaction surveys, healthcare systems and testing of web services see, respectively, Kenett, De Frenne, Tort-Martorell, and McCollin (2008), Kenett (2012, 2016, 2017) and Bai, Kenett, and Yu (2012), Kenett and Raanan (2010), Kenett and Salini (2011), Peterson and Kenett (2011). For examples of applications of BNs to education, banking, forensic and official statistics see Pietro et al. (2015), Di Zio, Sacco, Scanu, and Vicard (2005), Marella and Vicard (2013), Tarantola, Vicard, and Ntzoufras (2012), Vicard, Dawid, Mortera, and Lauritzen (2008). To fully specify a BN, and thus represent joint probability distributions, it is necessary to specify for each node X the probability distribution for X conditional upon X's parents.
Monitoring an airport check-in process by using Bayesian networks
2017, Transportation Research Part A: Policy and PracticeCitation Excerpt :In 2009, Renzi et al. highlighted the similarity between cause-and-effect diagrams and BNs by an application to the Italian public administration. BNs have also been proposed for managing information that originates from customer satisfaction surveys (Salini and Kenett, 2009; Kenett and Salini, 2012; Tarantola et al., 2012) and as a tool for benchmarking analysis (Guglielmetti and Musella, 2013). An evolution of BNs, namely object-oriented BNs, has been also used for handling complex managerial problems (Musella and Vicard, 2015).
Customer-oriented communication in retail and Net Promoter Score
2017, Journal of Retailing and Consumer ServicesMapping resource conflicts with probabilistic network models
2016, Journal of Cleaner ProductionCitation Excerpt :The main approaches to structural learning are the scoring and searching method (Cooper and Herskovits, 1992; Heckerman, 1995), which selects the model with the highest score, and the constraint-based methods (Spirtes et al., 2000), which use hypothesis tests for checking conditional independence. Given their simplicity both in visualizing and describing complex interrelations, BNs have been applied to different frameworks: customer satisfaction analysis (Salini and Kennett, 2009; Tarantola et al., 2012), joint monitoring of internal and external performance (Di Pietro et al., 2015) and benchmarking (Guglielmetti Mugion and Musella, 2013). The increasing use of BNs in several fields is mainly due to their ability to conduct a what-if analysis.
Risk assessment and risk management of violent reoffending among prisoners
2015, Expert Systems with ApplicationsCitation Excerpt :The implications are extended to the interventional modelling case in the sense that the BN demonstrates how actions are supported by the model, with respect to determining whether a prisoner’s risk of violence can be managed to acceptable levels after release on the basis of some causal intervention, such as treatment, therapy and/or medication. The outcome of this paper is in general agreement with many other studies that demonstrate decision support benefits, in various other domains, using probabilistic graphical models (Constantinou et al., 2012; Constantinou et al., 2013; de Melo & Sanchez, 2008; Fenton & Neil, 2011; Fenton & Neil, 2012; Lee & Park, 2010; Lee et al., 2009; Pourret et al., 2008; Tarantola et al., 2012; Di Pietro et al., 2015; Zhang et al., 2013; Perkusich, 2015). In terms of AI and decision support research, the problem addressed in this paper is typical of many critical decision-making scenarios (especially in medicine, forensics, and transport safety assessment); specifically decision-makers are seeking improved methods for prediction and risk assessment, but have either little relevant data, or have to rely on poorly structured data.