Churn in services – A bibliometric review

This article is distributed under the terms of the Creative Commons Atribution 4.0 Internacional License institutua enpresa Instituto de Economía Aplicada a la Empresa Management Letters / Cuadernos de Gestión journal homepage: http://www.ehu.eus/cuadernosdegestion/revista/es/ ISSN: 1131-6837 / e-ISSN: 1988-2157 Management Letters Cuadernos de Gestión Enpresa Institutua, UPV/EHU Conocimiento en Gestión/Management Knowledge Volume 22 / Number 2 (2022) • ISSN: 1131-6837 / e-ISSN: 1988-2157 http://www.ehu.eus/cuadernosdegestion/revista/es/ Churn in services – A bibliometric review Abandono en servicios - Una revisión bibliométrica Hugo Ribeiro*, Belém Barbosa a, b, c , António C. Moreira b, d, e, f , Ricardo Rodrigues e, g a University of Porto. School of Economics and Management – [email protected] –https://orcid.org/0000-0002-4057-360X b GOCVOPP - Research Unit on Governance, Competitiveness and Public Policies, University of Aveiro, Portugal c cef.UP – Center for Economics and Finance at UPorto, University of Porto, Portugal d Aveiro University. Department of Economics, Management, Industrial Engineering, and Tourism – [email protected] – https://orcid.org/0000-0002-6613-8796 e NECE-UBI - Research Center for Business Sciences, Universidade da Beira Interior, Portugal f INESCTEC—Institute for Systems and Computer Engineering, Technology and Science, Faculdade de Engenharia da Universidade do Porto, Portugal g Universidade da Beira Interior, Department of Business and Economics – [email protected] – https://orcid.org/0000-0001-6382-5147 * Corresponding author: Aveiro University. Department of Economics, Management, Industrial Engineering, and Tourism–[email protected]–https://orcid.org/00000002-0410-6430 ARTICLE INFO Received 01 June 2021, Accepted 29 December 2021 Available online 4 March 2022 DOI: 10.5295/cdg.211509hr JEL: M30, M31 ABSTRACT The purpose of this article is to identify the most impactful research on customer churn and to map the conceptual and intellectual structure of its field of study. Data were collected from the WoS database, comprising 338 articles published between 1995 and 2020. Several bibliometric techniques were applied, including analysis of co-words, co-citation, bibliographic coupling, and co-authorship networks. R software and the Bibliometrix/Biblioshiny package were used to perform the analyses. The results identify the most active and influential authors, articles, and journals on the topic. More specifically, through co-citations and bibliographic coupling, it was possible to map the oldest articles (retrospective analysis) and the current research front (prospective analysis). The retrospective analysis, based on co-citations, revealed that the foundations of this research field are constructs such as quality of service, satisfaction, loyalty, and changing behaviors. The prospective analysis, performed through bibliographic coupling, revealed that current research is embedded in predictive analysis, clusters, data mining, and algorithms. The results provide robust guidance for further investigation in this field. Keywords: Customer Churn, Bibliometric Analysis, Co-citation Analysis, Bibliographic Coupling, Science Mapping, Biblioshiny. RESUMEN El objetivo de este artículo es identificar las investigaciones más impactantes sobre la pérdida de clientes y trazar la estructura conceptual e intelectual de su campo de estudio. Los datos han sido recogidos de la base de datos WoS, que comprenden 338 artículos publicados entre 1995 y 2020. Varias técnicas bibliométricas fueron aplicadas, incluyendo el análisis de co-palabras, cocitaciones, acoplamiento bibliográfico y redes de coautoría. Para realizar los análisis se utilizaron el software R y el Bibliometrix/Biblioshiny. Los resultados identifican los autores, artículos y revistas más influyentes y activos sobre el tema. Más específicamente, a través de las cocitaciones y el acoplamiento bibliográfico, fue posible mapear los artículos más antiguos (análisis retrospectivo) y la investigación más actual (análisis prospectivo). El análisis retrospectivo, basado en las cocitaciones, reveló que los fundamentos de este campo de investigación son constructos como la calidad del servicio, la satisfacción, la lealtad y el cambio de comportamientos. El análisis prospectivo, realizado a través del acoplamiento bibliográfico, reveló que la investigación actual está inmersa en el análisis predictivo, los conglomerados, la minería de datos y los algoritmos. Los resultados proporcionan una sólida orientación para seguir investigando en este campo. Palabras clave: Churn de Clientes, Análisis Bibliométrico, Análisis de Cocitación, Acoplamiento Bibliográfico, Mapeo de la Ciencia, Biblioshiny. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 98 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues 1. INTRODUCTION Customer churn is one of the most challenging topics for managers in several sectors involving contracted services, where customers tend to change service providers repeatedly (Kumar et al., 2018). Also known as customer turnover, and customer defection or departure, customer churn refers to a customer’s decision to cease business with a service provider (Adebiyi etal., 2016; Eshghi etal., 2007; Mahajan etal., 2015; Prince & Greenstein, 2014), by switching to a new provider. There are several clear indications of the relevance of this topic. Being widely accepted that attracting new customers by driving them away from a competitor is generally more expensive than retaining current customers by meeting their real needs (Kyei & Bayoh, 2017), the literature further stresses that higher retention rates result in higher market share, resulting in higher revenues (Kyei & Bayoh, 2017). Particularly in markets reaching high levels of saturation and concentration, finding and retaining new customers is increasingly difficult and expensive (Carrizo-Moreira et al., 2017; Hadden et al., 2007; Moreiraetal., 2016). For that reason, companies such as banks, telecommunications and airlines, to name but a few, use customer churn or retention rate as a critical business metric (Amiri & Daume III, 2016). In this connection, the literature suggests that higher priority should be given to retaining the most valuable existing customers rather than gaining new ones (Carrizo-Moreira etal., 2017; Hadden etal., 2007; Moreira etal., 2016), which leads to a change in business paradigm, with a new emphasis on customer retention and relationship development, rather than looking mainly at customer acquisition. Understanding and preventing customer churn, particularly by profitable customers, is critical to business survival (Adebiyi etal., 2016), including identifying the customers who are most likely to switch service providers (Amin etal., 2019). The determinants of customer churn have been the focus of a significant stream of research, remaining central constructs in marketing activities (Schweidel etal., 2008). Keaveney (1995) was one of the first authors to study customer churn, and found that its main causes included price, service failure, and the company’s responses to service problems. Rajan (2017) added that customer turnover may be based on dissatisfaction, higher costs, low quality, lack of resources, and privacy concerns. Satisfaction stands out as a predictor of customer retention (Anderson etal., 1994; Eshghi, Haughton, & Topi, 2007), but it also causes customers to switch service providers (Becker etal., 2015), namely in response to positive word-ofmouth from competitors’ most satisfied customers (de Haan etal., 2015). Despite the steep increase in published articles on customer churn, no one has yet provided a summarized review of the scientific landscape, preventing a clear view of the state of the art. To the best of our knowledge, no work so far has focused on analyzing the development of scientific production on customer churn. In order to address this gap and present information on how churn has been addressed in the literature, this article uses a bibliometric analysis. Once a scientific discipline has reached a certain degree of maturity, it is common practice for researchers to focus their attention on the literature generated by the scientific community to conduct literature reviews to assess the state of the art (Ramos-Rodriguez & Ruiz-Navarro, 2004). Indeed, synthesizing the results of past research is one of the most critical tasks for advancing knowledge in a particular research topic (Zupic & Cater, 2015), namely by adopting bibliometric research methods to map the structure and development of scientific fields and disciplines (Zupic & Cater, 2015). Specifically, this technique examines how disciplines, fields, expertise, and individual documents and authors relate (Small, 1999; Zupic & Cater, 2015). Bibliometric methods adopt a quantitative approach to describe, evaluate, and monitor published research, determining its cognitive structure and evolution (Small, 1999). They follow a systematic, transparent and replicable review process (Zupic & Cater, 2015) that systematically represents the nature of specific scientific disciplines, highlighting research trends (Zhang etal., 2016). As such, they identify major research areas, providing researchers with a solid basis for positioning significant current contributions and detecting new avenues for future research (Ferreira, 2018). Following the bibliometric analysis approach, this article aims to identify the most impactful research on customer churn and to map the conceptual and intellectual structure of its field of study. It is intended through this bibliometric analysis to answer the following research questions: — What are the specific topics associated with customer churn research? — What is the intellectual structure of the field? — Who are the central, peripheral, or bridging researchers in this field? — What is the intellectual structure of recent/emerging literature? — What is the social structure of the field? This article makes several contributions to development of the literature on customer churn in services. Firstly, it describes the structure of the conceptual field through co-word analysis and maps the intellectual field through a co-citation analysis of authors, articles, and journals. Secondly, it identifies and organizes the most recurrent themes and cutting-edge research through a bibliographic coupling analysis, which shows how the topic is developing. Finally, it identifies the social structure of the research field. Hence, this article provides scholars with guidance for future research, by highlighting the most prominent contributions on the topic and by identifying the trends in this field of research. This approach reveals that the literature is varied and covers topics such as defection, retention, customer churn, and switching behavior, encompassing different and sometimes complementary concepts. It is also worth noting that customer experience, disappointment, desertion, customer encounter and satisfaction, lack of service quality and attributes are among the main determinants of churning. Finally, it is important to mention that if predictive models are used extensively to analyze churn, behavioral models are also greatly used to understand what leads customers to swap one service provider for another. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 99 2. BIBLIOMETRIC ANALYSIS OF THE LITERATURE Bibliometric methods (for example, co-citation analysis, bibliographic coupling, analysis of co-authors, among others) use bibliographic data from publication databases to construct structural images of scientific fields (Zupic & Cater, 2015), discovering their essence (Pritchard, 1969). Bibliometric methods have two primary uses: performance analysis and scientific mapping (Cobo et al., 2011). Performance analysis seeks to assess the performance of research and publication by individuals and institutions. Scientific mapping aims to reveal the structure and dynamics of scientific fields (Zupic & Cater, 2015). Different approaches have been applied to extract networks using different units of analysis (authors, documents, journals, and terms). Table1 shows the techniques that will be applied throughout this document. Table 1 Most common bibliometric techniques by the unit of analysis Bibliometric technique Unit of analysis used Kind of relation Bibliographic Coupling Author Author’s oeuvres Common references among author’s oeuvres  Document Document Common references among documents Journal Journal’s oeuvres Common references among journal’s oeuvres Coauthor Author Author’s name Authors’ cooccurrence Country Country from affiliation Countries’ cooccurrence Institution Institution from affiliation Institutions’ cooccurrence Co-citation Author Author’s reference Co-cited author  Document Reference Co-cited documents Journal Journal’s reference Co-cited journal Co-word Keyword, or term extracted from title, abstract, or document’s body Terms’ cooccurrence Source: Adapted from Cobo etal. (2011). 2.1. Research methodology and article selection Two types of research objectives can be defined by using the bibliography for scientific mapping: (1)to identify the knowledge base of a research topic or field and its intellectual structure; (2)to identify the main themes and trends (conceptual structure). The analysis carried out in this study serves to identify the conceptual and intellectual structures of the research field. To identify the existence of clusters of articles on customer churn, the most influential authors, their geographical origin, and authorship networks. To construct the bibliometric maps, the software used was Biblioshiny (Aria & Cuccurullo, 2017), from the R Core Team (Team, 2021) Bibliometrix package. R is an opensource programming language, creating a software ecosystem accessible to the whole community. Indeed, all resources are shared by the community, and all knowledgeable users can contribute to the development of different software packages. Biblioshiny is an R software package that provides a graphical environment for using the Bibliometrix package. The Bibliometrix package allows the analysis and mapping of bibliographic data. Regarding data collection, and to ensure that all relevant articles were considered in this study, we initially performed a search on the WoS1 database with the main keyword “customer churn” and restricting the search to articles in English published until 2020. WoS is by far the most common source of bibliographic data (Zupic & Cater, 2015), with the oldest and most comprehensive records of citation indexes (Ellegaard & Wallin, 2015). This database contains enough data for most bibliometric analyses. Data, including article title, article type, authors, their institutional affiliations, keywords, abstract, number of citations, journal name, publisher name and address, year of publication, volume, issue number, and a list of cited references are available for analysis (Zupic & Cater, 2015). This search identified 253articles, whose keywords and abstracts were extracted and subject to content analysis using NVIVO2 qualitative analysis software. This procedure allowed us to identify synonyms, alternative phrasing, and combinations of the word “churn” with other words that could help identify additional articles on the topic. Content analysis software such as NVIVO was considered helpful because it can explore word search and analyze word frequency. As a result, additional keywords were identified for this study, including “customer turnover”, “customer switching”, “churn management”, “churn factors”, and “customer defection”. The resulting set of keywords is presented in Table2. Table 2 Keywords used in the search Keywords customer churn customer turnover customer attrition customer rotation customer defection customer switching consumer switching behavior customer switching behavior stayers * switchers churn management churn determinants churn factors churn analysis defection management Source: Author’s own elaboration. 1 http://www.webofknowledge.com 2 https://www.qsrinternational.com/nvivo-qualitative-data-analysis-software/home Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 100 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues After identifying all the keywords, a new search was carried out in WoS, always using the Boolean operator “or” between them. The search was based on the articles’ title, abstract and keywords. Only journal articles were considered, because this type of publication is arguably considered most highly by academics and practitioners, due to representing most peer-reviewed and also the most citable contributions. Hence, conference proceedings, books, book reviews, or conference abstracts were not considered. In addition, only articles written in English and published up to 2020 were considered, returning 452articles. Data were extracted and analyzed in the first semester of 2021. Figure 1 Search flowchart used in this investigation Source: Author’s own elaboration. Figure1 presents the search flowchart, created using the tool developed by Haddaway and McGuinness (2021), describing the various stages of screening and the number of articles excluded from the analysis. Firstly, all abstracts were read and analyzed in detail. As a result, 114articles were excluded, either because they were outside the scope of this study or because they addressed other research topics unrelated to customer churn. Hence, only articles that had customer churn as the object of study were included. Consequently, 338articles were considered for this study. In relation to bibliometric methods, different choices were made regarding the network layout, data normalization, and the clustering algorithm for the analysis of co-words, co-citations, bibliographic coupling, and co-authorship networks. Table3 presents the choices made for each analysis typology, as presented in the next sections. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 101 Table 3 Network Layout, Normalization and Clustering Algorithm Analysis Number of Nodes Minimum edges Network Layout Normalization Clustering Algorithm Co-Word 50 2 Kamada & Kawai Association Walktrap Co-Citation & Bibliographic Coupling    Authors 50 2 Kamada & Kawai N/A Walktrap Articles 50 2 Kamada & Kawai N/A Walktrap Journals 50 2 Kamada & Kawai N/A Walktrap Coauthor    Authors 50 2 Fruchterman & Reingold Association Walktrap Country of Affiliation 50 2 Fruchterman & Reingold Association Walktrap Institution from affiliation 50 2 Fruchterman & Reingold Association Walktrap Source: Author’s own elaboration. The data were normalized using association strength as a similarity measure. According to the theoretical and empirical results of Eck and Waltman (2009), this measure was considered the most appropriate to normalize data co-occurrence. A clustering algorithm was used to divide the overall network into different sub-networks to perform community detection (Cobo etal., 2011). A community, or cluster, is generally defined as a subset of densely interconnected nodes relative to the rest of the network (Newman & Girvan, 2004). Walktrap, developed by Pons and Latapy (2005), was the algorithm used. As explained by its authors, Walktrap offers a measure of similarity between vertices based on random walks, with several important advantages: it captures the community structure of a network well, can be efficiently computed, and can be used in an agglomeration algorithm to compute efficiently the community structure of a network. This community detection algorithm was found to have one of the best results in identifying communities even for high values of mixing coefficient (Orman & Labatut, 2009). The next step was to select co-occurrences, involving network filtering using two as a minimum edge value reduction. Only edges with a strength greater than or equal to two are plotted. In all analyses, only the 50 most cited references were plotted. 2.2. Descriptive Analysis We started by performing a descriptive analysis of the database used, to summarize and explore the data. Table 4 shows the main statistics for the database. The period of analysis is from 1995 to 2020, presenting an annual growth rate of 17.84%. The 338articles on the database have an average citation of 33.24. The articles were written by 796 authors, with an average of 2.36authors per article. Table 4 Main information about data Description Articles 338 Period 1995 - 2020 Annual Percentage Growth Rate 17,84% Average citations per article 33,24 Authors 796 Authors Appearances 995 Authors of single-authored articles 34 Authors of multi-authored articles 762 Authors per Article 2,36 Co-Authors per Articles 2,94 Collaboration Index 2,53 Source: Author’s own elaboration. As shown in Figure2, the number of articles published before 2009 was relatively small, but publications increase after that period. 0 5 10 15 20 25 30 35 40 1990199520002005201020152020 Number of publish articles Year Figure 2 Publishing trend in the area of customer churn Source: Author’s own elaboration. Note: Number of published articles per year. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 102 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues The initial statistics show that the 338 articles were published in 177 different journals. The top 10journals were responsible for 110 articles, about 33% of the total. Table5 shows the leading journals where these articles were published. “Expert Systems with Applications” stands out, with 12% of all publications. Table 5 The top 10 publishing journals contributing to the area of customer churn Sources No. of Articles % of Articles Expert Systems with Applications 41 12 European Journal of Operational Research 9 3 European Journal of Marketing 8 2 Marketing Science 8 2 Journal of Marketing Research 7 2 Journal of Service Research 7 2 Telecommunications Policy 7 2 Decision Support Systems 6 2 Journal of The Academy of Marketing Science 6 2 International Journal of Advanced Computer Science and Applications 5 1 Source: Author’s own elaboration. Table6 presents the top 10 authors and the number of articles of which they are authors or coauthors. Van Den Poel D, Baesens B, and Coussement, K have the greatest number of articles. These three authors have their publications mostly in “Operations Research Management Science” and “Computer Science Artificial Intelligence”. Table 6 The top 10 contributing authors and number of articles Author Number of published articles Van Den Poel D 13 Baesens B 12 Coussement K 8 Verbeke W 6 Amin A 5 Anwar S 5 Liu Y 5 Martens D 5 Source: Author’s own elaboration. Table 7 shows the top10 most cited articles. Those with the highest number of citations are (1)Customer switching behaviour in service industries: An exploratory study (Keaveney, 1995); (2)Consumer switching costs: A typology, antecedents, and consequences (Burnham et al., 2003) and (3) Switching barriers and repurchase intentions in services (Jones et al., 2000). Table 7 Top 10-Most cited articles (Global) Article Total Global Citations (TC) TC per Year Keaveney Sm, 1995, J Mark 1154 42,74 Burnham Ta, 2003, J Acad Mark Sci 756 39,79 Jones Ma, 2000, J Retail 648 29,45 Bansal Hs, 2004, J Acad Mark Sci 427 23,72 Chen Py, 2002, Inf Syst Res 339 16,95 Aydin S, 2005, Eur J Market 260 15,29 Neslin Sa, 2006, J Mark Res 238 14,88 Keaveney Sm, 2001, J Acad Mark Sci 237 11,29 Wei Cp, 2002, Expert Syst Appl 198 9,90 Burez J, 2009, Expert Syst Appl 189 14,54 Source: Author’s own elaboration. It is also interesting to analyze the top 10 most cited articles locally: articles that received citations from documents contained only in the database studied, as shown in Table 8. The most cited articles are (1)New insights into churn prediction in the telecommunication sector: A profit-driven data mining approach (Verbeke etal., 2012); (2)Turning telecommunications call details to churn prediction: a data mining approach (Wei & Chiu, 2002); and (3)Customer attrition analysis for financial services using proportional hazard models (Van den Poel & Lariviere, 2004), with the exact quotes that (3)Customer base analysis: partial defection of behaviorally loyal clients in a non-contractual FMCG retail setting (Buckinx & Van den Poel, 2005). Table 8 Top 10-Most cited articles (Local) Article Total Local Citations (TC) Verbeke W, 2012, Eur J Oper Res 55 Wei Cp, 2002, Expert Syst Appl 54 Van Den Poel D, 2004, Eur J Oper Res 44 Buckinx W, 2005, Eur J Oper Res 44 Burez J, 2009, Expert Syst Appl 43 Tsai Cf, 2009, Expert Syst Appl 39 Verbeke W, 2011, Expert Syst Appl 39 Ahn Jh, 2006, Telecommun Policy 35 Burez J, 2007, Expert Syst Appl 34 Hadden J, 2007, Comput Oper Res 27 Source: Author’s own elaboration. Regarding the authors’ affiliation, Table9 shows the organizations contributing most, based on the number of articles published. Comparing this list with the list of the top10 authors, in Table6, the Catholic University of Leuven, the University of Ghent, and the University of Southampton are represented by the most prolific authors, Van den Poel, Dirk (Ghent University) and Baesens, Bart (University of Southampton and Catholic University of Leuven). Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 103 Table 9 The top 20 contributing organizations Organization Location Articles Katholieke Univ Leuven Belgium 31 University Ghent Belgium 21 University Southampton United Kingdom 16 Inst Management Sci Pakistan 10 University Catholique Lille France 10 University Tehran Iran 10 Deakin University Australia 8 Iqra National University Pakistan 8 University Coll Dublin Ireland 8 Georgia State University United States 7 Hong Kong Polytech University China 7 King Abdulaziz University Saudi Arabia 7 K.N. Toosi University Technology Iran 7 University Oviedo Spain 7 Utah State University United States 7 Columbia University United States 6 Helwan University Egypt 6 Kuhne Logistic University Germany 6 Lebanese American University Lebanon 6 National Central University Taiwan 6 Source: Author’s own elaboration. Figure3 shows the articles published by country. The intensity of the color is proportional to the number of publications. In general, the geographical dispersion indicates that research and practice regarding customer churn have attracted worldwide attention. Figure 3 Scientific Production by Country Source: Author’s own elaboration-Biblioshiny output. Note: The shade of blue represents the production of articles per country. The darker the color, the greater the production. Table10 shows the top10 countries by publications, considering the corresponding author, with China, the United States, and Belgium accounting for 41% of the total. Table 10 Top 10 Most productive countries (based on first author’s affiliation) Country Number of published articles % of Articles China 53 14 USA 51 13 Belgium 25 6 Korea 19 5 Pakistan 16 4 India 15 4 Iran 14 4 Australia 11 3 France 10 3 Germany 9 2 Source: Author’s own elaboration. The most frequently used words/phrases were also analyzed. The analysis was performed for the words/phrases defined by the original authors (Author Keywords) and the words/phrases defined automatically by a computerized algorithm (KeywordsPlus). These words or phrases that appear defined by the algorithm are frequently present in the titles of the references of an article and not necessarily in the title of the article or the author’s keywords (Garfield & Sher, 1993). The KeywordsPlus algorithm can capture the content of articles with greater depth and variety (Garfield & Sher, 1993). It is as effective as the words defined by the authors in terms of bibliometric analysis investigating the knowledge structure of scientific areas. Still, it is less comprehensive in representing the content of an article (Zhang etal., 2016). The top20 authors’ words can be seen in Table11, and the top20 of the words generated by the KeywordsPlus algorithm in Table12. Table 11 Top 20-Most frequent Author Keywords Author Keywords (DE) No. of Articles Author Keywords (DE) No. of Articles Churn Prediction 66 Customer Churn Prediction 14 Data Mining 47 Customer Loyalty 14 Customer Churn 45 Telecommunications 12 Classification 29 Churn Analysis 11 Customer Relationship Management 28 Customer Defection 11 Customer Retention 26 Logistic Regression 11 Churn 22 Satisfaction 11 Machine Learning 19 Telecommunication 11 Customer 18 Marketing 10 Customer Satisfaction 15 Prediction 10 Source: Author’s own elaboration. Regarding the words defined by the authors, “Customer Churn”, “Customer Defection” and “Churn Analysis” should be read with due contextualization since they were part of the set of terms used to build the research. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 104 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues Table 12 Top 20-Most frequent Keywords-Plus Keywords-Plus (ID) No. of Articles Keywords-Plus (ID) No. of Articles Satisfaction 74 Determinants 28 Retention 63 Defection 27 Model 60 Quality 25 Loyalty 52 Management 24 Behavior 39 Customer Churn 23 Services 38 Selection 22 Models 37 Algorithm 20 Prediction 37 Attrition 20 Classification 33 Customer Satisfaction 18 Impact 32 Dynamic-Model 18 Source: Author’s own elaboration. From analysis of the two tables, terms linked to data science —“Churn Prediction”, “Data Mining”, “Classification”, “Machine Learning”, “Prediction”, “Model”, “Algorithm” and “Selection”— stand out, representing the prominence of data analysis and data mining in investigating customer churn. It is also interesting to note that customer retention, satisfaction and loyalty, and prediction also appear at the top of the most referenced words, either by the original authors or by the KeywordPlus algorithm, revealing the importance of these constructs in investigating customer churn. Figures4 and 5 show the evolutionary trend of the top10 words/phrases. Concerning the authors’ words, there is a growing trend in “Churn Prediction”, “Machine Learning”, and “Customer Churn”, once again reinforcing the previous observations. Concerning KeywordPlus words/phrases, and therefore with a broader spectrum, we observe adjacent research fields with significant growth, “satisfaction”, “retention” and “behavior”, but also the same terms that were observed in the authors’ words, related to data analysis, “Model” and “Prediction”. Figure 4 Timeline Word Growth (Author’s Keywords) Source: Author’s own elaboration-Biblioshiny output. Note: Number of occurrences of the author keywords over time. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 105 Figure 5 Timeline Word Growth (KeywordsPlus) Source: Author’s own elaboration-Biblioshiny output. Note: Number of keyword plus occurrences over time. 2.3. Conceptual structure of the field The conceptual structure allows understanding of what is being said in the research field, the main themes, and trends (Aria & Cuccurullo, 2017). It is often used to understand the topics covered by researchers and identify the most important and most recent issues. The most frequent co-word analysis was used for the conceptual structure and to understand what was being analyzed in the research field. A. Co-word analysis Co-word analysis is a content analysis technique that uses the most important words or keywords in documents, to establish relationships and build a conceptual structure of a research field, and to identify the main concepts analyzed in a given field (Callon etal., 1991; Callon etal., 1983). It can be applied to the keywords, abstracts, or texts in their entirety (Aria & Cuccurullo, 2017). The unit of analysis is usually a keyword or term extracted from the title, abstract, or body of the document (Aria & Cuccurullo, 2017). In scientific mapping, a network graph is used to represent co-occurrences between bibliographic metadata. The graph is made up of nodes or points (each node is a word) connected by lines. The size of each node is proportional to the occurrence of the item, and the size of the edges of the lines is proportional to their co-occurrence. The colors represent the groups to which each word belongs. Co-occurrences can be normalized using similarity measures to obtain similarities between the data (Cobo etal., 2011). Figure6 shows the network diagram resulting from the authors’ co-words analysis. Two major groups/clusters of words can be distinguished. The first group deals with customer churn based on predictive analysis, feature selection, clustering, data mining, and algorithms such as decision trees, random forests, logistic regression, and support vector machine. Also appearing in this group are words such as big data, business intelligence, and telecom. The second group of words is related to customer retention, satisfaction, loyalty, customer relationship management, service quality, change intentions, and behavior. Relational marketing, trust, and switching costs also figure in this group. These two clusters show that the research field is essentially sustained by two streams of research, the most recurrent of which uses predictive methods. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 112 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues 2.5. Intellectual structure of the field - Bibliographic Coupling The intellectual structure reveals how an author’s work influences a determined scientific community. Co-citation analysis and bibliographic coupling have been used to analyze the intellectual structure of a field of scientific research (Cobo etal., 2011). Bibliographic coupling answers questions such as; What is the intellectual structure of recent/emerging literature? (Zupic & Cater, 2015). Bibliographic coupling is particularly suitable for analyzing the research front of a research topic or field (Zupic & Cater, 2015). The concept of a research front is used to describe current scientific articles that cite the knowledge base publications, previously described as the set of articles most cited by current research. Two articles are linked bibliographically if at least one cited source appears in the bibliographies or reference lists of both articles (Kessler, 1963). Bibliographic coupling describes the extent to which two articles are related because both refer to the same article (Ferreira, 2018). This analysis technique can also be applied to authors and journals. The author’s bibliographic coupling aims to discover co-author relationships between authors who cite the same references, while bibliographic coupling of journals aims to discover journals that cite the same references (Cobo etal., 2011). For bibliographic coupling analysis, the 50 most cited authors, articles, and journals were selected. A. Authors Bibliographic coupling of authors is a method of mapping active authors, which can give a realistic view of the current state of research (Zhao & Strotmann, 2008). It can also analyze the social structure of a particular research field. Figure11 presents the resulting network diagram. The centrality measures of the co-authorship network were calculated. Based on betweenness centrality and closeness centrality, as explained previously, the authors with the greatest measure of proximity are: (1)Van Den Poel D; (2)Coussement K; and (3)Hsieh YC. Table17 shows the top 10 authors regarding these two metrics (the complete list or the database itself can be obtained from the corresponding author upon request). We can conclude that these authors are the main constituents of the intellectual structure of the research field. They are the authors of the research front. Table 17 Top 10 Authors with high Betweenness and Closeness (BC) Authors Closeness Authors Betweenness Van Den Poel D 0.242 Mahajan V 0.012 Coussement K 0.241 Zhang Y 0.012 Hsieh YC 0.240 Kim D 0.010 Mahajan V 0.239 Van Den Poel D 0.009 Gerpott T 0.239 Gerpott TJ 0.008 Stakhovych S 0.239 Coussement K 0.008 Lee YS 0.238 Polo Y 0.006 De Bock KW 0.238 Javier Sese F 0.006 Ahmadi N 0.238 Gupta S 0.006 Ewing M 0.238 Ahmadi N 0.006 Source: Author’s own elaboration. Figure 11 Bibliographic coupling of authors Source: Author’s own elaboration-Bibliometrix output. Note: The size of the circle indicates an item’s weight, the lines indicate the links between the items, the distance between the items shows their relationship, and the different colors indicate the clusters. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 113 B. Articles As mentioned previously, bibliographic coupling describes the extent to which two articles are related by both citing the same article (Ferreira, 2018). Figure12 shows the network diagram resulting from bibliographic coupling of the articles. Centrality measures for the network and the respective nodes were calculated. The articles with the greatest centrality measure are: (1)Improving customer attrition prediction by integrating emotions from client/company interaction emails and evaluating multiple classifiers (Coussement & Van den Poel, 2009); (2)A new hybrid classification algorithm for customer churn prediction based on logistic regression and decision trees (De Caigny etal., 2018); and (3)Regaining drifting mobile communication customers: Predicting the odds of success of winback efforts with competing risks regression (Gerpott & Ahmadi, 2015). Figure 12 Bibliographic Coupling of Articles Source: Author’s own elaboration-Bibliometrix output. Note: The size of the circle indicates an item’s weight, the lines indicate the links between the items, the distance between the items shows their relationship, and the different colors indicate the clusters. Table18 shows the top10 articles concerning their degree of closeness centrality, that is, the articles most similar to each other. Table19 shows the articles with a higher degree of betweenness centrality: the articles representing a bridge between the different research streams. The articles in both tables constitute the research front on customer churn. The majority of articles come after 2009, with 90% of the articles in the tables belonging to this period. More than half the articles are from 2015 onwards. This was also observed in the descriptive analysis, in the evolution of article publication over time. The research front is related mainly to predictive analysis. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 114 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues Table 18 Top 10 Articles with high Closeness (BC) Authors Article Closeness Coussement K, 2009 Improving customer attrition prediction by integrating emotions from client/company interaction emails and evaluating multiple classifiers 0.309 De Caigny, 2018 A new hybrid classification algorithm for customer churn prediction based on logistic regression and decision trees 0.308 Gerpott TJ, 2015 Regaining drifting mobile communication customers: Predicting the odds of success of winback efforts with competing risks regression 0.306 Coussement K, 2014 Improving customer retention management through cost-sensitive learning 0.306 Mahajan V, 2015 Review of data mining techniques for churn prediction in telecom 0.306 Ahn J, 2020 A survey on churn analysis in various business domains 0.306 Jahromi AT 2014 Managing B2B customer churn, retention and profitability 0.305 Gerpott TJ, 2015 Who is (not) convinced to withdraw a contract termination announcement? - A discriminant analysis of mobile communications customers in Germany 0.304 Verbeke W, 2012 New insights into churn prediction in the telecommunication sector: A profit driven data mining approach 0.304 Buckinx W, 2005 Customer base analysis: Partial defection of behaviourally loyal clients in a non-contractual FMCG retail setting 0.300 Source: Author’s own elaboration. Table 19 Top 10 Articles with high Betweenness (BC) Authors Article Betweenness Ahn Y, 2020 Customer attrition analysis in the securities industry: A large-scale field study In Korea 0.014 Mahajan V, 2015 Review of data mining techniques for churn prediction in telecom 0.014 Coussement K, 2009 Improving customer attrition prediction by integrating emotions from client/company interaction emails and evaluating multiple classifiers 0.013 Gerpott TJ, 2015 New insights into churn prediction in the telecommunication sector: A profit driven data mining approach 0.013 Benedek G, 2014 The importance of social embeddedness: churn models at mobile providers 0.011 Ahn J, 2020 A survey on churn analysis in various business domains 0.011 Al-Mashraie M, 2020 Customer switching behavior analysis in the telecommunication industry 0.010 De Caigny, 2018 A new hybrid classification algorithm for customer churn prediction 0.009 Polo Y, 2009 How to Make Switching Costly: The role of marketing and relationship characteristics 0.009 Hadden J 2007 Computer-assisted customer churn management: State-of-the-art and future trends 0.008 Source: Author’s own elaboration. C. Journals The bibliographic coupling of journals seeks to study the relationship between common references among the journals’ publications (Cobo etal., 2011). Figure 13 presents the network diagram resulting from the bibliographic coupling of journals. Centrality measures were calculated for the co-authorship network. Regarding closeness centrality, the journals with the highest value are (1)Expert Systems with Applications, (2)Telecommunications Policy, (3)European Journal of Marketing. Expert Systems with Applications has the highest number of articles in the sample under study, at 12%. Regarding betweenness centrality, the journals with the highest values are (1)International Journal of Bank Marketing; (2) Expert Systems with Applications; and (3) European Journal of Operational Research. Table20 shows the top10 journals in terms of centrality measures. From the research field of the journals, we can conclude that the research front has been primarily published by journals whose focus is on intelligent systems, technologies, applications, intelligence, and data science. Also noteworthy are two journals, which are more sector-oriented, with publications on telecommunications and banking. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 115 Figure 13 Bibliographic Coupling of Sources Source: Author’s own elaboration-Bibliometrix output. Note: The size of the circle indicates an item’s weight, the lines indicate the links between the items, the distance between the items shows their relationship, and the different colors indicate the clusters. Table 20 Top 10 Journals with high Betweenness and Closeness (BC) Journal Closeness Journal Betweenness Expert Systems With Applications 0.326 International Journal of Bank Marketing 0.035 Telecommunications Policy 0.321 Expert Systems with Applications 0.030 European Journal of Marketing 0.321 European Journal of Operational Research 0.025 European Journal of Operational Research 0.320 Decision Sciences 0.025 Industrial Marketing Management 0.319 European Journal of Marketing 0.021 International Journal of Bank Marketing 0.317 Journal of Business Research 0.016 Journal of Business Research 0.317 Telecommunications Policy 0.015 Journal of Service Research 0.317 Journal of Service Research 0.015 Marketing Science 0.315 Marketing Science 0.015 Information Systems and e-Business Management 0.310 Industrial Marketing Management 0.014 Source: Author’s own elaboration. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 116 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues 2.6. Co-author analysis Co-author analysis looks at authors and their affiliations to study social structure and collaboration networks (Peters & Vanraan, 1991). It is particularly well suited to studying research questions involving scientific collaboration (Zupic & Cater, 2015). The most commonly used method to study social structure is the co-author network. However, it is also possible to use information about authors’ geographical location and institutional affiliations to examine collaboration at the level of institutions and countries (Zupic & Cater, 2015). A. Author Figure14 shows the co-authorship network graph. There are 10 clusters, and three stand out with a higher number of co-authors. The first was led by Baesens B and Verbeke W, the second by Van Den Poel D and Coussement K, and the third by AminA and AnwarS. These authors also have the highest number of articles in the sample studied. Table 21 presents the author networks, ordering them by PageRank. For co-authoring networks, PageRank gives greater weight to authors who collaborate with different authors and those who collaborate with few authors but do so frequently (Yan & Ding, 2011). Figure 14 Co-authorship analysis Source: Author’s own elaboration-Biblioshiny output. Note: The rectangles denote author nodes and are labeled by the author’s last name and initials. The node size corresponds to the number of publications, and the color of the different clusters. Table 21 Articles of each co-authorship cluster: PageRank measure Cluster 1 PageRank Cluster 2 PageRank Van den Poel D 0.037 Stakhovych S 0.026 Coussement K 0.028 Ewing M 0.026 Migueis Vl 0.026   Camanho As 0.026   Falcao E 0.026   Cunha J 0.026   De Bock Kw 0.024   De Caigny A 0.019   Cluster 3 PageRank Cluster 4 PageRank Becker Ju 0.026 Amin A 0.040 Spann M 0.026 Anwar S 0.040   Shah B 0.021   Nawaz M 0.021   Hussain A 0.021   Al-Obeidat F 0.016 Cluster 5 PageRank Cluster 6 PageRank Baesens B 0.063 Kim Ys 0.026 Verbeke W 0.036 Lee H 0.026 Vanthienen J 0.026   Martens D 0.023   Oskarsdottir M 0.019   Stripling E 0.009   Zhu B 0.009   Cluster 7 PageRank Cluster 8 PageRank Guillen M 0.026 Li H 0.026 Nielsen Jp 0.026 Liu Y 0.026 Cluster 9 PageRank Cluster 10 PageRank Raza B 0.026 Khan Y 0.026 Malik Ak 0.026 Shafiq S 0.026   Ahmed S 0.026   Safwan N 0.026 Source: Author’s own elaboration. This analysis also shows that the clusters of authors with the greatest scientific collaboration are essentially those identified as mainly responsible for the research front through the bibliographic coupling analysis, meaning there is high collaboration on the research front. Van Den Poel D and Coussement K stand out with the highest centrality measure, meaning they can reach other authors in the network via a shorter path. B. Country of affiliation Regarding co-authorship in the author’s country of affiliation, Figure15 presents five clusters, two of which have a higher number of co-authors. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 117 Figure 15 Co-authorship analysis (Country) Source: Author’s own elaboration-Biblioshiny output. Note: The nodes represent countries and links connect countries in the form of co-authorships. Bibliometrix software attributes a different color to each cluster. The first cluster comprises the United States and China with a higher PageRank, and a second made up of the United Kingdom and Belgium with a high PageRank. We recall that China, the United States, and Belgium are responsible for 41% of all publications. Table22 presents the country networks, sorting them by PageRank. Table 22 Countries of each co-authorship cluster: PageRank measure Cluster 1 PageRank Cluster 2 PageRank United Kingdom 0.121 Spain 0.036 Belgium 0.106 Denmark 0.023 France 0.042  Portugal 0.016  Cluster 3 PageRank Cluster 4 PageRank Germany 0.041 Pakistan 0.0637 Switzerland 0.026 United Arab Emirates 0.0214  Saudi Arabia 0.0245 Cluster 5 PageRank  USA 0.170  China 0.148  Korea 0.057  India 0.031  Netherlands 0.020  Canada 0.019  Australia 0.016  Turkey 0.016  Source: Author’s own elaboration. Figure16 shows the collaborative map between countries, where the frequencies between China and the United States and between the United Kingdom and Belgium stand out, as seen earlier. Figure 16 Country CollaborationMap Source: Author’s own elaboration-Biblioshiny output. Note: The blue color on the map represents research cooperation among countries. The darker the color, the higher the country collaboration. The scale of cooperation is represented through the thickness of the line. C. Institution from affiliation Finally, concerning the authors’ affiliated institutions, the co-authorship network reveals one prominent cluster, with the University of Southampton and the Catholic University of Leuven at the head, affiliated universities of Baesens B, who is also the author with the second highest number of articles on the database. Figure17 presents the graph of the co-authorship network by educational institution. Figure 17 Co-authorship analysis (Institution) Source: Author’s own elaboration-Biblioshiny output. Note: The nodes represent institutions and are labeled by the institution name. The node size corresponds to the number of occurrences of co-authorships publications, and links connect institutions in the form of co-authorships. Bibliometrix software attributes different colors to each cluster. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 118 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues Table23 shows the networks of institutions by cluster, ordered by PageRank. Table 23 Institutions of each co-authorship cluster: PageRank measure Cluster 1 PageRank Cluster 2 PageRank Katholieke University Leuven 0.121 Inst Management Sci 0.084 University Southampton 0.108 Zayed University 0.039 University of Antwerp 0.035 Taibah University 0.050 Vrije Universiteit Brussel 0.026 University Stirling 0.050 Sichuan University 0.026  University Ghent 0.017  Cluster 3 PageRank Cluster 4 PageRank Columbia University 0.056 Monash University 0.003 University of Pennsylvania 0.056 Deakin University 0.003 Cluster 5 PageRank  Chinese Academy of Sciences 0.056  University Nebraska 0.056  Source: Author’s own elaboration. 3. CONCLUSIONS AND DIRECTIONS FOR FUTURE RESEARCH As markets become increasingly saturated, academic researchers and companies have recognized it is essential to identify the customers most likely to switch to another service provider (Keaveney & Parthasarathy, 2001). The intention here was to present a structured review of customer churn, using bibliometric techniques to analyze the research field’s intellectual and conceptual structure. As far as we know, a bibliometric analysis has never been carried out to identify analytically and objectively the most influential studies and authors, as well as the emerging research clusters, so we believe that this study advances knowledge about the field of studies analyzed. This study synthesizes insights from a considerable amount of relevant literature on customer churn. The data show that research on customer churn has been published more and more, thus demonstrating the growing importance of this field of research. As for the most influential outlets for the dissemination of research about customer churn, Expert Systems with Applications is the journal of choice. Amongst the most prolific authors, we highlight Van Den Poel, Baesens B and Coussement K, which explains, in part, why Belgium is one of the countries that have contributed most to developing and disseminating research on customer churn. Interestingly, when we focus our analysis on the most cited references, the work by Keaveney S.M ranks very highly, suggesting the relevance of this author’s research in this area of knowledge. However, when one analyzes the global and local citations, one realizes that Verbeke and Burez stand out among the top-10 most local-cited articles, Burez being in the last place in the top-10 gobal-cited articles. One possible explanation why they are not among the top-10 most globally cited articles is that their publications are from 2011 and 2012 (Verbeke) and from 2007 and 2009 (Burez). In order to complement the descriptive analysis and obtain further insights into the customer churn literature, this article analyzed the conceptual and intellectual structure of the research field. Determination of the intellectual structure and the research front of the scientific domains are essential not only for research but also to elaborate policies and practices (Aria & Cuccurullo, 2017). The importance of having “a conceptual and intellectual map” is undeniable for the construction of a holistic view of a field of studies. The bibliometric analysis (1995-2000) carried out in this study allows mapping and synthesizing the relationships between authors, articles, and fundamental journals in the field of customer churn. Regarding the first research question proposed for this article, which was about the specific topics associated with customer churn research, we performed a co-word analysis. We observed in the first instance that there are two different angles, two major lines of investigation, which are complementary. One, in which the researchers’ objectives are to understand what leads to customer churn and to define essential churn factors, such as satisfaction, service quality and service attributes (e.g., Athanassopoulos, 2000; Keaveney, 1995; Reichheld & Sasser, 1990; Rust & Zahorik, 1993). The second, where researchers focus on improving customer churn forecasting models to boost predictive performance (e.g., Verbeke etal., 2012; Verbeke etal., 2011; Wei& Chiu, 2002). These predictive studies typically apply modeling techniques, such as artificial neural networks, decision trees, and random forests, to large samples of subscribers. Complementarily, and through the construction of a thematic map3, churn prediction was found to be the most important theme in the field of investigation. This theme relates to different concepts such as “classification”, “machine learning” and “telecommunication”, among others. Since telecommunications is an area suitable for this type of investigation, future research on customer turnover is recommended, using not predictive models, but behavioral models, with antecedents focusing on satisfaction, quality of service, switching costs, and socio-demographic data, to name just a few. However, despite these findings, churn seems to be an unresolved issue as customer experience, encounter, disappointment and interaction are continuously present in the services provided. Moreover, beyond the relational aspects of situational interaction, there is a lack of knowledge on how cumulative experience and customer encounters influence churn. Regarding the intellectual structure of the field, approached by the second research question, two techniques of bibliometric analysis were used, co-citation analysis and bibliographic coupling. Firstly, through analyzing articles, journals and authors, we sought to map older works, identifying the intellectual structure at the base of the field of studies through analysis of co-ci3 Thematic mapping consists of a word co-occurrence network analysis to define what science talks about in a research field, main themes, and trends. Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 119 tations. Secondly, we mapped the current front of the research field through bibliographic coupling. Co-citation analysis revealed that the authors with the greatest closeness and betweenness centrality are Reichheld FF, Keaveney SM, Bolton RN, Rust RT, and Ganesh J, that is, these are the primary authors most cited by current research, and they are the foundations of current research. The article “Customer switching behavior in service industries: An exploratory study” by Keaveney (1995), which in the intellectual structure of the research field emerges as the article with the greatest measure of closeness centrality, is one of the main founding articles of the research field and is also the most cited article globally. It was the first article to present a model explaining customer churn in service industries considering a possible number of causal factors and their interrelationships These data tackle the third research question proposed by this article, regarding the central, peripheral, or bridging researchers in this field. Regarding cutting-edge research and the intellectual structure of emerging literature (fourth research question), bibliographic coupling revealed that the principal authors are Coussement K and Van den Poel D, with the most prominent article being by these authors, Improving Customer Attrition Prediction by Integrating Emotions from Client/Company Interaction Emails and Evaluating Multiple Classifiers (Coussement & Van den Poel, 2009). Hence, predictive methods are found to be at the cutting edge of customer churn research. Finally, in relation to the social structure of the research field (fifth research question), the United States and China are the main collaborating countries. This collaboration should be extended to other clusters of countries, since customer turnover is transversal to all countries, and a more comprehensive view of cross-cultural management is clearly necessary. Throughout the study, and in all the analyses performed, no references were found to the research field about customer experience. A good customer experience tends to reduce significantly the propensity to switch to another brand (Sirapracha & Tocquer, 2012). Experiences can be seen as the “impressions” that remain in customers’ minds, as the result of a holistic encounter with an offer or object (Iglesiasetal., 2011), culminating in satisfaction or disappointment and desertion or abandonment (Meyer & Schwager, 2007). A positive experience can promote an emotional bond between the brand and its customers, which, in turn, increases customer loyalty (Gentile etal., 2007). One recommendation for future investigation is to study the relationship between the customer experience and customer churn, the determinants and their impact on customer churn. We share the same question referred to by Lemon and Verhoef (2016) who ask if customer experience can explain the customer’s behavior and the company’s performance. It is also important to address how disconnected the relation between customer satisfaction and involvement with churn is. Once again, we emphasize the need to approach churn addressing the behavioral side, not only with predictive methods but also addressing how emotional drivers influence behavioral responses leading to churn, both in relational or transactional encounters. Moreover, there is lack of references to how competitors’ action influences churning behavior. Although this investigation contributes to knowledge in this area, being the first attempt to map the research field systematically, several limitations and future research opportunities are worth mentioning through a bibliometric analysis. Data collection was carried out exclusively on the “Web of Science” database and, consequently, articles and other types of documents indexed by other platforms were not covered. This decision was because the software used for bibliometric analysis is still unable to merge the references cited correctly. In the future, other databases (e.g., Scopus) should be analyzed together with WoS. Only scientific articles were considered, excluding books, conference proceedings, editorial material, and others, so it will be of interest to examine other publications not included in the sample, to complement the results obtained. Moreover, the focus of this article was on bibliometric analysis and the intellectual structure of churn. As such, a literature review based on content analysis is mandatory to explore the intricacies relating consumer behavior, customer satisfaction, customer loyalty, churn prediction, customer churn and their emotional drivers. 4. REFERENCES Adebiyi, S. O., Oyatoye, E. O., & Amole, B. B. (2016). Relevant drivers for customers` churn and retention decision in the Nigerian mobile telecommunication industry. Journal of Competitiveness, 6(3), 5267. https://doi.org/10.7441/joc.2016.03.04 Ahn, J. H., Han, S. P., & Lee, Y. S. (2006). Customer churn analysis: Churn determinants and mediation effects of partial defection in the Korean mobile telecommunications service industry. Telecommunications Policy, 30(10-11), 552-568. https://doi.org/10.1016/j. telpol.2006.09.006 Ahn, J., Hwang, J., Kim, D., Choi, H., & Kang, S. (2020). A Survey on churn analysis in various business domains. IEEE Access, 8, 220816220839. https://doi.org/10.1109/access.2020.3042657 Al-Mashraie, M., Chung, S. H., & Jeon, H. W. (2020). Customer switching behavior analysis in the telecommunication industry via push-pullmooring framework: A machine learning approach. Computers & Industrial Engineering, 144. https://doi.org/10.1016/j.cie.2020.106476 Amin, A., Shah, B., Khattak, A. M., Moreira, F. J. L., Ali, G., Rocha, A., & Anwar, S. (2019). Cross-company customer churn prediction in telecommunication: A comparison of data transformation methods. International Journal of Information Management, 46, 304-319. https://doi.org/10.1016/j.ijinfomgt.2018.08.015 Amiri, H., & Daume III, H. (2016). Short text representation for detecting churn in microblogs. Paper presented at the Thirtieth AAAI Conference on Artificial Intelligence. Anderson, E. W., Fornell, C., & Lehmann, D. R. (1994). Customer satisfaction, market share, and profitability - Findings from Sweden. Journal of Marketing, 58(3), 53-66. https://doi.org/10.1177/002224299405800304 Aria, M., & Cuccurullo, C. (2017). Bibliometrix: An R-tool for comprehensive science mapping analysis. Journal of Informetrics, 11(4), 959-975. https://doi.org/10.1016/j.joi.2017.08.007 Athanassopoulos, A. D. (2000). Customer satisfaction cues to support market segmentation and explain switching behavior. Journal of Business Research, 47(3), 191-207. https://doi.org/10.1016/s01482963(98)00060-5 Aydin, S., & Ozer, G. (2005). The analysis of antecedents of customer loyalty in the Turkish mobile telecommunication market. European Journal of Marketing, 39(7-6), 910-925. https://doi. org/10.1108/03090560510601833 Bansal, H. S., Irving, P. G., & Taylor, S. F. (2004). A three-component model of customer commitment to service providers. Journal of the Academy of Marketing Science, 32(3), 234-250. https://doi. org/10.1177/0092070304263332 Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 120 Hugo Ribeiro, Belém Barbosa, António C. Moreira, Ricardo Rodrigues Becker, J. U., Spann, M., & Schulze, T. (2015). Implications of minimum contract durations on customer retention. Marketing Letters, 26(4), 579-592. https://doi.org/10.1007/s11002-014-9293-2 Benedek, G., Lubloy, A., & Vastag, G. (2014). The Importance of Social Embeddedness: Churn Models at Mobile Providers. Decision Sciences, 45(1), 175-201. https://doi.org/10.1111/deci.12057 Bolton, R. N. (1998). A dynamic model of the duration of the customer's relationship with a continuous service provider: The role of satisfaction. Marketing Science, 17(1), 45-65. https://doi.org/10.1287/mksc.17.1.45 Buckinx, W., & Van den Poel, D. (2005). Customer base analysis: Partial defection of behaviourally loyal clients in a non-contractual FMCG retail setting. European Journal of Operational Research, 164(1), 252-268. https://doi.org/10.1016/j.ejor.2003.12.010 Burez, J., & Van den Poel, D. (2007). CRM at a pay-TV company: Using analytical models to reduce customer attrition by targeted marketing for subscription services. Expert Systems with Applications, 32(2), 277-288. https://doi.org/10.1016/j.eswa.2005.11.037 Burez, J., & Van den Poel, D. (2009). Handling class imbalance in customer churn prediction. Expert Systems with Applications, 36(3), 4626-4636. https://doi.org/10.1016/j.eswa.2008.05.027 Burnham, T. A., Frels, J. K., & Mahajan, V. (2003). Consumer switching costs: A typology, antecedents, and consequences. Journal of the Academy of Marketing Science, 31(2), 109-126. https://doi. org/10.1177/0092070302250897 Callon, M., Courtial, J. P., & Laville, F. (1991). Co-word analysis as a tool for describing the network of interactions between basic and technological research - The case of polymer chemistry. Scientometrics, 22(1), 155-205. https://doi.org/10.1007/bf02019280 Callon, M., Courtial, J. P., Turner, W. A., & Bauin, S. (1983). From translations to problematic networks - An introduction to co-word analysis. Social Science Information Sur Les Sciences Sociales, 22(2), 191235. https://doi.org/10.1177/053901883022002003 Carrizo-Moreira, A., Freitas-da Silva, P. M., & Ferreira-Moutinho, V. M. (2017). The effects of brand experiences on quality, satisfaction and loyalty: An empirical study in the telecommunications multiple-play service market. Innovar, 27(64), 23-38. https://doi.org/ 10.15446/innovar.v27n64.62366 Chen, P. Y., & Hitt, L. M. (2002). Measuring switching costs and the determinants of customer retention in Internet-enabled businesses: A study of the Online brokerage industry. Information Systems Research, 13(3), 255-274. https://doi.org/10.1287/isre.13.3.255.78 Cobo, M. J., Lopez-Herrera, A. G., Herrera-Viedma, E., & Herrera, F. (2011). Science mapping software tools: review, analysis, and cooperative study among tools. Journal of the American Society for Information Science and Technology, 62(7), 1382-1402. https://doi. org/10.1002/asi.21525 Coussement, K. (2014). Improving customer retention management through cost-sensitive learning. European Journal of Marketing, 48(3-4), 477-495. https://doi.org/10.1108/ejm-03-2012-0180 Coussement, K., & Van den Poel, D. (2009). Improving customer attrition prediction by integrating emotions from client/company interaction emails and evaluating multiple classifiers. Expert Systems with Applications, 36(3), 6127-6134. https://doi.org/10.1016/j. eswa.2008.07.021 De Caigny, A., Coussement, K., & De Bock, K. W. (2018). A new hybrid classification algorithm for customer churn prediction based on logistic regression and decision trees. European Journal of Operational Research, 269(2), 760-772. https://doi.org/10.1016/j. ejor.2018.02.009 de Haan, E., Verhoef, P. C., & Wiesel, T. (2015). The predictive ability of different customer feedback metrics for retention. International Journal of Research in Marketing, 32(2), 195-206. https://doi. org/10.1016/j.ijresmar.2015.02.004 Eck, N. J. v., & Waltman, L. (2009). How to normalize co-occurrence data? An analysis of some well‐known similarity measures. Journal of the American society for information science and technology, 60(8), 1635-1651. https://doi.org/10.1002/asi.21075 Ellegaard, O., & Wallin, J. A. (2015). The bibliometric analysis of scholarly production: How great is the impact? Scientometrics, 105(3), 1809-1831. https://doi.org/10.1007/s11192-015-1645-z Eshghi, A., Haughton, D., & Topi, H. (2007). Determinants of customer loyalty in the wireless telecommunications industry. Telecommunications Policy, 31(2), 93-106. https://doi.org/10.1016/j. telpol.2006.12.005 Ferreira, F. A. F. (2018). Mapping the field of arts-based management: Bibliographic coupling and co-citation analyses. Journal of Business Research, 85, 348-357. https://doi.org/10.1016/j.jbusres.2017.03.026 Fischbach, K., Putzke, J., & Schoder, D. (2011). Co-authorship networks in electronic markets research. Electronic Markets, 21(1), 19-40. https://doi.org/10.1007/s12525-011-0051-5 Ganesh, J., Arnold, M. J., & Reynolds, K. E. (2000). Understanding the customer base of service providers: An examination of the differences between switchers and stayers. Journal of Marketing, 64(3), 65-87. https://doi.org/10.1509/jmkg.64.3.65.18028 Garfield, E., & Sher, I. H. (1993). KEYWORDS-PLUS(TM) - Algorithmic derivative indexing. Journal of the American Society for Information Science, 44(5), 298-299. https://doi.org/10.1002/(sici)10974571(199306)44:5<298::aid-asi5>3.0.co;2-a Gentile, C., Spiller, N., & Noci, G. (2007). How to sustain the customer experience:: An overview of experience components that co-create value with the customer. European management journal, 25(5), 395410. https://doi.org/ /10.1016/j.emj.2007.08.005 Gerpott, T. J., & Ahmadi, N. (2015). Regaining drifting mobile communication customers: Predicting the odds of success of winback efforts with competing risks regression. Expert Systems with Applications, 42(21), 7917-7928. https://doi.org/10.1016/j.eswa.2015.05.011 Haddaway, N. R. A. U. P. C. C., & McGuinness, L. A. (2021). PRISMA2020: R package and ShinyApp for producing PRISMA 2020 compliant flow diagrams (Version 0.0.2): Zenodo. Retrieved from http://doi.org/10.5281/zenodo.5082518 Hadden, J., Tiwari, A., Roy, R., & Ruta, D. (2007). Computer assisted customer churn management: State-of-the-art and future trends. Computers & Operations Research, 34(10), 2902-2917. https://doi. org/10.1016/j.cor.2005.11.007 Iglesias, O., Singh, J. J., & Batista-Foguet, J. M. (2011). The role of brand experience and affective commitment in determining brand loyalty. Journal of Brand Management, 18(8), 570-582. https://doi. org/10.1057/bm.2010.58 Jahromi, A. T., Stakhovych, S., & Ewing, M. (2014). Managing B2B customer churn, retention and profitability. Industrial Marketing Management, 43(7), 1258-1268. https://doi.org/10.1016/j.indmarman.2014.06.016 Jones, M. A., Mothersbaugh, D. L., & Beatty, S. E. (2000). Switching barriers and repurchase intentions in services. Journal of Retailing, 76(2), 259-274. https://doi.org/10.1016/s0022-4359(00)00024-5 Keaveney, S. M. (1995). Customer switching behavior in-service industries - An exploratory-study. Journal of Marketing, 59(2), 71-82. https://doi.org/10.2307/1252074 Keaveney, S. M., & Parthasarathy, M. (2001). Customer switching behavior in online services: An exploratory study of the role of selected attitudinal, behavioral, and demographic factors. Journal of the Academy of Marketing Science, 29(4), 374-390. https://doi. org/10.1177/03079450094225 Kessler, M. M. (1963). Bibliographic coupling between scientific papers. American Documentation, 14(1), 10-25. https://doi.org/10.1002/ asi.5090140103 Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121 Churn in services – A bibliometric review 121 Kumar, V., Leszkiewicz, A., & Herbst, A. (2018). Are you back for good or still shopping around? Investigating customers’ repeat churn behavior. Journal of Marketing Research, 55(2), 208-225. https://doi. org/10.1509/jmr.16.0623 Kyei, D. A., & Bayoh, A. T. M. (2017). Innovation and customer retention in the Ghanaian telecommunication industry. International Journal of Innovation, 5(2), 171-183. https://doi.org/10.5585/iji.v5i2.154 Lemmens, A., & Croux, C. (2006). Bagging and boosting classification trees to predict churn. Journal of Marketing Research, 43(2), 276-286. https://doi.org/10.1509/jmkr.43.2.276 Lemon, K. N., & Verhoef, P. C. (2016). Understanding customer experience throughout the customer journey. Journal of Marketing, 80(6), 69-96. https://doi.org/10.1509/jm.15.0420 Mahajan, V., Misra, R., & Mahajan, R. (2015). Review of data mining techniques for churn prediction in telecom. Journal of Information and Organizational Sciences, 39(2), 183-197. Meyer, C., & Schwager, A. (2007). Understanding customer experience. Harvard Business Review, 85(2), 116-26,157. Moreira, A. C., Silva, P., & Moutinho, V. (2016). Differences between stayers, switchers, and heavy switchers: A study in the telecommunications service market. Marketing Intelligence & Planning, 34(6), 843-862. https://doi.org/10.1108/MIP-07-2015-0128 Mozer, M. C., Wolniewicz, R., Grimes, D. B., Johnson, E., & Kaushansky, H. (2000). Predicting subscriber dissatisfaction and improving retention in the wireless telecommunications industry. Ieee Transactions on Neural Networks, 11(3), 690-696. https://doi.org/10.1109/72.846740 Neslin, S. A., Gupta, S., Kamakura, W., Lu, J. X., & Mason, C. H. (2006). Defection detection: Measuring and understanding the predictive accuracy of customer churn models. Journal of Marketing Research, 43(2), 204-211. https://doi.org/10.1509/jmkr.43.2.204 Newman, M. E., & Girvan, M. (2004). Finding and evaluating community structure in networks. Physical review E, 69(2), 026113. https:// doi.org/10.1103/physreve.69.026113 Orman, G. K., & Labatut, V. (2009). A comparison of community detection algorithms on artificial networks. Paper presented at the International conference on discovery science. https://doi.org/10.1007/978-3642-04747-3_20 Peters, H. P. F., & Vanraan, A. F. J. (1991). Structuring scientific activities by coauthor analysis - An exercise on a university-faculty level. Scientometrics, 20(1), 235-255. https://doi.org/10.1007/bf02018157 Polo, Y., & Sese, F. J. (2009). How to Make Switching Costly The Role of Marketing and Relationship Characteristics. Journal of Service Research, 12(2), 119-137. https://doi.org/10.1177/1094670509335771 Pons, P., & Latapy, M. (2005). Computing communities in large networks using random walks. Paper presented at the International symposium on computer and information sciences. https://doi.org/10.1007/11569596_31 Prince, J., & Greenstein, S. (2014). Does service bundling reduce churn? Journal of Economics & Management Strategy, 23(4), 839-875. https://doi.org/10.1111/jems.12073 Pritchard, A. (1969). Statistical bibliography or bibliometrics. Journal of documentation, 25(4), 348-349. Rajan Sachdeva, R. R. S. (2017). Evaluating prediction of customer churn behavior based on artificial bee colony algorithm. International Journal of Engineering and Computer Science, 6(1). https:// doi.org/10.18535/ijecs/v6i1.32 Ramos-Rodriguez, A. R., & Ruiz-Navarro, J. (2004). Changes in the intellectual structure of strategic management research: A bibliometric study of the Strategic Management Journal, 1980-2000. Strategic Management Journal, 25(10), 981-1004. https://doi.org/10.1002/smj.397 Reichheld, F. F., & Sasser, W. E. (1990). Zero defections - Quality comes to services. Harvard Business Review, 68(5), 105-111. Rust, R. T., & Zahorik, A. J. (1993). Customer satisfaction, customer retention, and market share. Journal of Retailing, 69(2), 193-215. https://doi.org/10.1016/0022-4359(93)90003-2 Sirapracha, J., & Tocquer, G. (2012). Customer experience, brand image and customer loyalty in telecommunication services. Paper presented at the Int Conf Econ Bus Mark Manag. Small, H. (1973). Co‐citation in the scientific literature: A new measure of the relationship between two documents. Journal of the American Society for information Science, 24(4), 265-269. https://doi. org/10.1002/asi.4630240406 Small, H. (1999). Visualizing science by citation mapping. Journal of the American Society for Information Science, 50(9), 799-813. https://doi. org/10.1002/(sici)1097-4571(1999)50:9<799::aid-asi9>3.0.co;2-g Team, R. C. (2021). R: A language and environment for statistical computing. Tsai, C. F., & Lu, Y. H. (2009). Customer churn prediction by hybrid neural networks. Expert Systems with Applications, 36(10), 1254712553. https://doi.org/10.1016/j.eswa.2009.05.032 Ullah, I., Raza, B., Malik, A. K., Imran, M., Ul Islam, S., & Kim, S. W. (2019). A churn prediction model using random forest: Analysis of machine learning techniques for churn prediction and factor identification in telecom sector. Ieee Access, 7, 60134-60149. https://doi. org/10.1109/access.2019.2914999 Van den Poel, D., & Lariviere, B. (2004). Customer attrition analysis for financial services using proportional hazard models. European Journal of Operational Research, 157(1), 196-217. https://doi. org/10.1016/s0377-2217(03)00069-9 van Eck, N. J., & Waltman, L. (2010). Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics, 84(2), 523-538. https://doi.org/10.1007/s11192-009-0146-3 Verbeke, W., Dejaeger, K., Martens, D., Hur, J., & Baesens, B. (2012). New insights into churn prediction in the telecommunication sector: A profit driven data mining approach. European Journal of Operational Research, 218(1), 211-229. https://doi.org/10.1016/j.ejor.2011.09.031 Verbeke, W., Martens, D., Mues, C., & Baesens, B. (2011). Building comprehensible customer churn prediction models with advanced rule induction techniques. Expert Systems with Applications, 38(3), 2354-2364. https://doi.org/10.1016/j.eswa.2010.08.023 Wei, C. P., & Chiu, I. T. (2002). Turning telecommunications call details to churn prediction: a data mining approach. Expert Systems with Applications, 23(2), 103-112. https://doi.org/10.1016/s09574174(02)00030-1 Yan, E. J., & Ding, Y. (2011). Discovering author impact: A PageRank perspective. Information Processing & Management, 47(1), 125-134. https://doi.org/10.1016/j.ipm.2010.05.002 Zeithaml, V. A., Berry, L. L., & Parasuraman, A. (1996). The behavioral consequences of service quality. Journal of Marketing, 60(2), 31-46. https://doi.org/10.2307/1251929 Zhang, J., Yu, Q., Zheng, F. S., Long, C., Lu, Z. X., & Duan, Z. G. (2016). Comparing keywords plus of WOS and author keywords: A case study of patient adherence research. Journal of the Association for Information Science and Technology, 67(4), 967-972. https://doi. org/10.1002/asi.23437 Zhao, D. Z., & Strotmann, A. (2008). Evolution of research activities and intellectual influences in information science 1996-2005: Introducing author bibliographic-coupling analysis. Journal of the American Society for Information Science and Technology, 59(13), 2070-2086. https://doi.org/10.1002/asi.20910 Zupic, I., & Cater, T. (2015). Bibliometric methods in management and organization. Organizational Research Methods, 18(3), 429-472. https://doi.org/10.1177/1094428114562629 Management Letters / Cuadernos de Gestión 22/2 (2022) 97-121