6 _15-17_ - Adebiyi International Journal of DETERMINANTS OF CUSTOMERS` CHURN TELECOMMUNICATION INDUSTRY: AN ANALYTIC HIERARCHY PROCESS APPROACH University of Agriculture Abeokuta (FUNAAB), Ogun State. Nigeria Received: May 31, 2015 This paper describes the application of Analytic Hierarchy Process (AHP) for unraveling customers’ motivation for churn of telecommunication network in Nigeria. By identifying, modeling and measuring of customers` churn motivations across four mobile telecommunication service providers in Nigeria. AHP was used to design a hierarchical model of seven criteria for customers` churning of network and investigates the relative priorities of the criteria through a pairwise comparison. The questionnaire were admin through convenient sampling to 480 mobile telecommunication customers and was completed and returned by 438 mobile phone subscribers in Lagos state, Nigeria, but only 408 copies were useful for the analysis of this study. The result shows that six have weight above 10% in their individual contribution to motivating customer churn behavior in the Nigeria telecommunication industry. The criterion has the highest weight of 18.81% relative to the effectively supported modeling and decision for both the individual and the intelligible framework of decision emerging market like Nigeria. Keywords: Customers` Churn; marketing strategies. International Journal of Economic Behavior, vol 5, n. 1, ppInternational Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 DETERMINANTS OF CUSTOMERS` CHURN DECISION IN THE NIGERIA TELECOMMUNICATION INDUSTRY: AN ANALYTIC HIERARCHY PROCESS APPROACH SULAIMON OLANREWAJU ADEBIYI University of Agriculture Abeokuta (FUNAAB), Ogun State. Nigeria EMMANUEL OLATEJU OYATOYE University of Lagos, Akoka, Lagos. Nigeria BILQIS BOLANLE AMOLE University of Lagos, Akoka, Lagos. Nigeria Accepted: August 24, 2015 Online Published: Abstract This paper describes the application of Analytic Hierarchy Process (AHP) for unraveling customers’ motivation for churn of telecommunication network in Nigeria. By identifying, modeling and measuring of customers` churn motivations across four mobile ommunication service providers in Nigeria. AHP was used to design a hierarchical model of seven criteria for customers` churning of network and investigates the relative priorities of the criteria through a pairwise comparison. The questionnaire were admin through convenient sampling to 480 mobile telecommunication customers and was completed and returned by 438 mobile phone subscribers in Lagos state, Nigeria, but only 408 copies were useful for the analysis of this study. The result shows that six out of the seven criteria have weight above 10% in their individual contribution to motivating customer churn in the Nigeria telecommunication industry. The inefficient data/ internet plan criterion has the highest weight of 18.81% relative to the churn decision. Thus, effectively supported modeling and analyzing subscribers` motivation toward good marketing decision for both the individual and the organization. It helps in developing an analytic and intelligible framework of decision-making on complex problem of customer churn in an . Customers` Churn; AHP; churn decision; telecommunication International Journal of Economic Behavior, vol 5, n. 1, pp. 81- 81 14, 201 DETERMINANTS OF CUSTOMERS` CHURN DECISION IN THE NIGERIA TELECOMMUNICATION INDUSTRY: AN ANALYTIC HIERARCHY PROCESS APPROACH SULAIMON OLANREWAJU ADEBIYI University of Agriculture Abeokuta (FUNAAB), Ogun State. Nigeria EMMANUEL OLATEJU OYATOYE Akoka, Lagos. Nigeria BILQIS BOLANLE AMOLE Akoka, Lagos. Nigeria Online Published: October 7, 2015 This paper describes the application of Analytic Hierarchy Process (AHP) for unraveling customers’ motivation for churn of telecommunication network in Nigeria. By identifying, modeling and measuring of customers` churn motivations across four mobile ommunication service providers in Nigeria. AHP was used to design a hierarchical model of seven criteria for customers` churning of network and investigates the relative priorities of the criteria through a pairwise comparison. The questionnaire were administered through convenient sampling to 480 mobile telecommunication customers and was completed and returned by 438 mobile phone subscribers in Lagos state, Nigeria, but only 408 copies out of the seven criteria have weight above 10% in their individual contribution to motivating customer churn inefficient data/ internet plan churn decision. Thus, AHP subscribers` motivation toward good marketing . It helps in developing an analytic and complex problem of customer churn in an telecommunication; motivation; 104, 2015 82 1. Introduction Telecommunication is one of the prime support services needed for rapid growth of any developing economy (Arora, 2013). In Nigeria, the telecommunication industry received a boost in 2001 with the deregulation (removal of monopoly rights, especially enjoyed by state- owned telecommunication networks) of the sector. The deregulation allows for private Global System of Mobile communication (GSM) service providers to come into the country to support the state owned telecommunication company popularly known as Nigeria Telecommunication (NITEL). Consequently, two private companies ECONET of Zimbabwe and MTN of South Africa were licensed by the National Communications Commission (NCC) to operate GSM in Nigeria. These two companies alone increased the mobile telephone lines from 300,000 in 2001 to 1,660,000 in 2002. In 2003, another private-owned service provider, GLOBACOM, entered the market with its mobile service called the Glo- mobile (OECD report, 2003/2004). For this therefore, the number of mobile-phone subscriptions increased from less than one million in 2001 to over 133 million by November 2014 (NCC reports, 2014). Thus, mobile telecommunication popularly called GSM was well received by most Nigerians who have been battling with inefficiency of the only one service provider (NITEL) prior to the year 2001. After thirteen years of deregulation and globalisation of Nigeria telecommunication industry, Nigeria currently has five major GSM service providers namely: Airtel, MTN, Globalcom, Etisalat and the less functioning M-tel. Therefore, there are currently options for customers to choose from among multiple service providers and actively exercise their rights of switching from one service provider to another. This right possessed by customer to switch from service provider to another is known as churn. As it applies to this study, churn is explicitly defined as the act of leaving or abandoning a service provider for another. As a result, it become a constant practice among customers/subscribers of mobile telecommunication service providers in Nigeria. Furthermore, Mobile Number Portability (MNP) implementation in April 2013 lends credence to subscribers’ right to churn by switching their network providers at will on an average of every six months without losing their original number. The advent of cheap “china phones” that combine two to three or even up to four subscribers` identity module or subscriber identification module (SIM) cards in a phone, made it possible for most Nigerians to keep an average of two to three SIM cards of different service providers for various reasons that can be best described by each subscriber. As mobile telecommunication service now becomes essential to our daily life, it is observed that subscribers have learnt to keep their mobile phone (s) at arm’s length at all times, thus, predicting the reasons why mobile churn seems essential from the perspective of mobile service providers. Since, unguided movement of mobile phone subscribers from one service provider to another is one of the major phenomenon that can make or mar the survival and the profitability of any Mobile telecommunication firm. Thus, the need for research driven policy and strategies on customers` churn decision in the growing Nigeria mobile telecommunication industry cannot be over emphasised. Although, previous studies appear to have focused on churn prediction itself, using different statistical tools like data mining (Kolajo and Adeyemo, 2012), survival analysis (Van den Poel and Larivie`re, 2004), logistic regression (Kim and Yoon, 2004; Burez and Van den Poel, 2007), most of these seem to have failed to capture churn motivations, which may serve as good indicator for real churn forecasting. This is a major gap that this study intends to fill with Analytic Hierarchy process method. This study is important because, the annual churn rate of Nigeria mobile telecommunication according to NCC reports, have grown exponentially from 2 percent in International Journal of 2001 to 48.4 percent in 2007 (Pyramid Research, 2010). This became worrisome the consequences of customer loss of revenue, problem of referral, survival in the light fierce competition, and the MNP implementation in April 2013. The on customer churn issues worthwhile. Moreover, most prior studies used database of the firms (service providers) w switching behaviour and their causes. (AHP), based questionnaire could termination of contract with servi customers who churn for various reasons that will be supplied through pr collection, it will help stakeholders to link service attributes to subscribers’ decisions and make necessary inference. This analysis. It should also be noted that customer churn is a notorious problem for most industries, since customers are the lifeblood of any organization. 2. Literature Review AHP is a popular tool for decision released, many individuals and groups in various fields have used the AHP because of its user-friendly interface for multi from a decision-maker’s judgments, degree of importance of each alternative is quantified in the AHP. This procedure identifies not only the most important alternative decision-maker (Oyatoye, Adebiyi and Amole the AHP to analyse the decision respondents’ preferences/motivation use it in this study. The Analytic Hierarchy Process (AHP) is arguably multi-criteria method. It has firm theoretical underpinnings and has been used s help people make better decisions in a wide variety of complex circumstances (Golden, and Harker, 1989; Vaidya and Kumar 2006). A main strength of the AHP is that it is both methodologically sound and user design characteristics. The AHP frames a decision as a hierarchy, an organis already familiar with and easy to explain to those who are not. All inputs consist of comparisons between just two are generally considered to be one of the best ways to elicit judgments from people (Reynolds and Jolly, 1980). The output is easy to understand because it is based on simple scales derived from the pairwise compariso judgments being made which both checks the reliability of the analysis and reduces the chance of making a procedural mistake. of this approach is that it organiz provides a structured relatively simple solution to the decision by breaking a problem down in steps, to the smaller and smaller, one is able to connect, through simple paired comparison judgments, the small to the large (Lu, Madu, Kuei and Winokur, 1994 In sum, there are three major concepts behind the AHP Mathematical and logical reasoning for arriving at the decision is the strength of the AHP. It helps to analyse the decision problem on a logical footing and International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 in 2007 (Pyramid Research, 2010). This became worrisome the consequences of customers` churn on mobile service providers` profit, cost of loss of revenue, problem of referral, survival in the light fierce competition, and the MNP implementation in April 2013. These and other worrisome statistics trend to on customer churn issues worthwhile. most prior studies used the data or information available in the internal s (service providers) which mask detailed information about customers switching behaviour and their causes. While data, through Analytical Hierarchy Process questionnaire could provide more information on the reasons for customer termination of contract with service provider and help to make a better distinction between customers who churn for various reasons that will be supplied through primary source of data collection, it will help stakeholders to link service attributes to subscribers’ decisions and This in a considerable way, formed the basis for data used in AHP analysis. It should also be noted that customer churn is a notorious problem for most are the lifeblood of any organization. for decision-making developed by Saaty (1977, 1980). Since it was released, many individuals and groups in various fields have used the AHP because of its friendly interface for multi-criteria decision-making (Vargas, 1990). I maker’s judgments, known as pairwise comparisons, are aggregated, and the degree of importance of each alternative is quantified in the AHP. This procedure identifies not only the most important alternative, but also the preference for all alternatives for each , Adebiyi and Amole, 2015; Crawford and Williams, 1985). e the decision-making process, therefore, results in a precise clarification of /motivation for the alternatives (Sato, 2007). Thus, the decision to erarchy Process (AHP) is arguably the most well known and widely used criteria method. It has firm theoretical underpinnings and has been used s help people make better decisions in a wide variety of complex circumstances (Golden, 1989; Vaidya and Kumar 2006). A main strength of the AHP is that it is both methodologically sound and user-friendly. Its ease of use is due to a unique combination of ision as a hierarchy, an organisational framework many people are already familiar with and easy to explain to those who are not. All inputs consist of comparisons between just two decision elements at a time; pairwise comparisons like these are generally considered to be one of the best ways to elicit judgments from people (Reynolds 1980). The output is easy to understand because it is based on simple scales derived the pairwise comparisons. Besides, there is in-built measure of consistency in judgments being made which both checks the reliability of the analysis and reduces the chance of making a procedural mistake. According to Saaty and Kearns (1991), the stre this approach is that it organizes tangible and intangible factors in systematic way, and provides a structured relatively simple solution to the decision-making problems. by breaking a problem down into a logical fashion from the larger, descending in gradual steps, to the smaller and smaller, one is able to connect, through simple paired comparison judgments, the small to the large (Lu, Madu, Kuei and Winokur, 1994; Stephen, 1984 , there are three major concepts behind the AHP, namely: The AHP is analytic athematical and logical reasoning for arriving at the decision is the strength of the AHP. It the decision problem on a logical footing and assists on converting decision 83 14, 201 in 2007 (Pyramid Research, 2010). This became worrisome, considering churn on mobile service providers` profit, cost of operations, loss of revenue, problem of referral, survival in the light fierce competition, and the MNP trend to make a research information available in the internal hich mask detailed information about customers Analytical Hierarchy Process on the reasons for customers` ce provider and help to make a better distinction between imary source of data collection, it will help stakeholders to link service attributes to subscribers’ decisions and formed the basis for data used in AHP analysis. It should also be noted that customer churn is a notorious problem for most making developed by Saaty (1977, 1980). Since it was released, many individuals and groups in various fields have used the AHP because of its making (Vargas, 1990). In the AHP, data pairwise comparisons, are aggregated, and the degree of importance of each alternative is quantified in the AHP. This procedure identifies ference for all alternatives for each Crawford and Williams, 1985). Using making process, therefore, results in a precise clarification of Thus, the decision to the most well known and widely used criteria method. It has firm theoretical underpinnings and has been used successfully to help people make better decisions in a wide variety of complex circumstances (Golden, Wasli 1989; Vaidya and Kumar 2006). A main strength of the AHP is that it is both due to a unique combination of ational framework many people are already familiar with and easy to explain to those who are not. All inputs consist of decision elements at a time; pairwise comparisons like these are generally considered to be one of the best ways to elicit judgments from people (Reynolds 1980). The output is easy to understand because it is based on simple scales derived built measure of consistency in the judgments being made which both checks the reliability of the analysis and reduces the and Kearns (1991), the strength es tangible and intangible factors in systematic way, and making problems. In addition, er, descending in gradual steps, to the smaller and smaller, one is able to connect, through simple paired comparison ; Stephen, 1984). The AHP is analytic: athematical and logical reasoning for arriving at the decision is the strength of the AHP. It n converting decision- 84 makers’ intuition and gut feelings into numbers which can be openly questioned and be explained by others. The AHP structures the problem as a hierarchy: Hierarchic decomposition comes naturally to human beings. Reducing the complex problem into sub-problems to be tackled one at a time is the fundamental way that human decision-making have worked. Evidence from psychological studies suggests that human beings can compare 7 ± 2 things at a time. Hence, to deal with a large and complex decision making problem such as customer churn motivation, it is essential to break it down as a hierarchy. The AHP allows that. The AHP defines a process for decision-making: Formal processes for decision-making are the need of the hour. Decisions, especially collective ones, need to evolve. A process is required that will incorporate the decision-maker’s inputs, revisions and learning, and communicate them to others so as to reach a collective decision. The AHP has been created to formalize the process and place it on a scientific footing. It also helps in aiding the natural decision-making process. AHP approach is selected in this research to provide an effective tool to the stakeholders for unraveling the motivation for customer churn and measuring the churn drivers in service industry, which in turn, will make insightful contributions to the business world, marketing and operations research literature. Therefore, the AHP should be applied to churn decisions of subscribers. 3. Methodology Churning network provider in order to choose the most proffered telecommunication network provider is a complex problem requiring a multi-criteria decision analysis technique, so as to consider the factors holistically (criteria and alternatives) that motive customer churning behaviour. The Analytical hierarchy process was employed in this study in order to structure and simplify the complex problem, bring it to a condition, which is more easily understood. More importantly, the criteria weights and scores are based on pairwise comparisons of criteria and alternatives, respectively, using a ratio scale of measurement. In order to achieve the objectives of this study, survey research designed was found most appropriate and suitable due to the fact that surveys inquiry is about people`s attitude, lifestyles, behaviour, perception and problems (Leary, 2001). The survey is in two stages, first is the observation experiment through focus group discussion (FGD), consisting of fifteen subscribers in four groups who discussed extensively on the drivers of customer` churn in the telecommunication industry. The result of the FGD was used in the second stage for building the hierarchical model and drawing an AHP based questionnaires. The questionnaires were designed to enable each subscribers` to compare the relative importance of each driver to subscribers` decision to churn a network provider for the criteria, while the alternatives at the third level were equally pairwise to determine the relative importance of sub-criteria to each driver (criteria) of churn in the Nigeria telecommunication market. Eighty (80) copies of the questionnaires were administered using convenient sample in each of the six locations within Lagos state, corresponding to the administrative division of Lagos. The area includes Yaba, Ikorodu, Epe Badagry and Ikeja. Out of the total of 480 questionnaire administered, four hundred and eight (408) were properly filled and found suitable for the AHP analysis. The AHP analysis was done using Microsoft Excel software by calculating the weight of the criteria and alternatives. The process of analysis using AHP method involved two stages as follows (Taylor III, 2002): i. First Stage: build hierarchical model for Motivation for customer churn decision in the Nigerian telecommunication industry: (a) Establishing the pair-wise comparison matrix for each decision alternative for each criterion; (b) Synthesization; (c) International Journal of Establishing the pair-wise comparison matrix for each criteria normalized matrix; (e) values for each decision alternative; according to the values that have been acquired in the previous stage. ii. Second Stage: Test of Consistency method, the result of the selection process must be tested for consi consistency was carried out CI = [λmax – n]/(n – 1) where λmax=Σwici After acquiring Consistency Index (CI), th (CR) by using the formula CR = CI/RI where n is the number of items compared; W column; CR is the consistency ratio; CI is the consistency index; and RI consistency index. The Random Consistency Index (RI) can be observed in Table 1 as follows: Table1 – Random Index N 1 2 3 4 5 6 R.I. 0 0 0.58 0.90 1.12 1.25 Adapted from Saaty, (2000) If CR ≥ 10%, the data acquired is inconsistent, otherwise (CR < 10%) the data acquired is consistent. The results obtained from the above procedure discussion section of this paper. AHP Customer Churn model Level 1 The Goal: Determinant of custo industry. Level 2 The Criteria: Unwanted calls/SMS; dispute in billing; mobile number portability (MNP); frequency of promotions/bonuses by competitors; poor inter/intra connectivity; inefficient data plan; and poor complaint management. Level 3 The Alternatives: alternatives, the alternative of unwanted calls/SMS are and irrelevant calls from the network operator charge and charges for service(s) not rendered Easiness of MNP and competition brought by MNP are the alternatives of MNP as a driver for customer churn while frequenc on calls/SMS/data plan by competitor and poor/dubious promos of present operator as it alternatives. Poor intra/inter connectivity has its alternativ calls/sending SMS to same network and that of International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 wise comparison matrix for each criteria; (d) (e) Establishing the preference vector; (f) Calcul values for each decision alternative; and (g) Determining the rank of alternatives according to the values that have been acquired in the previous stage. Second Stage: Test of Consistency: After analyzing the data by using the AHP result of the selection process must be tested for consi carried out using the following formulas and the table 1) After acquiring Consistency Index (CI), the next step is calculating Consistency Ratio RI here n is the number of items compared; Wi is the weight; Ci is the sum along the column; CR is the consistency ratio; CI is the consistency index; and RI The Random Consistency Index (RI) can be observed in Table 1 as follows: 4 5 6 7 8 9 10 11 12 0.90 1.12 1.25 1.32 1.41 1.45 1.49 1.54 1.48 1.56 1.57 1.59 acquired is inconsistent, otherwise (CR < 10%) the data acquired is obtained from the above procedure were reported this paper. Determinant of customer churn decision in Nigerian telecommunication Level 2 The Criteria: Unwanted calls/SMS; dispute in billing; mobile number portability (MNP); frequency of promotions/bonuses by competitors; poor inter/intra connectivity; and poor complaint management. Level 3 The Alternatives: The components of each customer churn drivers formed the alternatives, the alternative of unwanted calls/SMS are frequently receiving irrelevant texts e network operator. For dispute in tariff; short d charges for service(s) not rendered by network operator are the alternatives Easiness of MNP and competition brought by MNP are the alternatives of MNP as a driver frequency of promotion has competitors’ frequent promos, bonuses on calls/SMS/data plan by competitor and poor/dubious promos of present operator as it Poor intra/inter connectivity has its alternatives as difficulties in making same network and that of other network while inefficient data p 85 14, 201 (d) Establishing the Calculating overall Determining the rank of alternatives according to the values that have been acquired in the previous stage. After analyzing the data by using the AHP result of the selection process must be tested for consistency. The test of s and the table 1. (3.1) e next step is calculating Consistency Ratio RI (3.2) is the sum along the column; CR is the consistency ratio; CI is the consistency index; and RI is the random The Random Consistency Index (RI) can be observed in Table 1 as follows: 9 10 11 12 13 14 15 1.45 1.49 1.54 1.48 1.56 1.57 1.59 acquired is inconsistent, otherwise (CR < 10%) the data acquired is were reported under results and in Nigerian telecommunication Level 2 The Criteria: Unwanted calls/SMS; dispute in billing; mobile number portability (MNP); frequency of promotions/bonuses by competitors; poor inter/intra connectivity; customer churn drivers formed the frequently receiving irrelevant texts short-change in service by network operator are the alternatives. Easiness of MNP and competition brought by MNP are the alternatives of MNP as a driver competitors’ frequent promos, bonuses on calls/SMS/data plan by competitor and poor/dubious promos of present operator as its es as difficulties in making inefficient data plan has 86 sufficient with high cost and insufficient with low cost data plan as its alternatives. The last driver (poor complaint management), has the following as its alternative; service agent non- willing to resolve customer challenges, non-response and service agent being elusive. Thus, the hierarchical model is presented in Figure 1. 4. Results and discussion 4.1 Respondents reduced matrices for customer churn decision The values found in the last column of table 2 denoted by weight, also known as eigenvector, have a direct physical meaning in AHP. They determine the participation or weight of those criteria relative to the total results of the goal. Considering the criteria stated for the determinant of customer churn in the Nigeria telecommunication industry, inefficient data/ internet plan criterion has a weight of 18.81% relative to the total goal, which states the determinant of customer churn. A positive evaluation on this factor contributes approximately twice more than a positive evaluation on the mobile number portability implementation criterion (9.73%). Following the procedure of AHP, there is need to check for data inconsistencies. The main objective is to capture enough information that will help to determine whether the customers have been consistent in their choices. The inconsistency index is based on maximum lambda, which is calculated by summing the product of each element in the eigenvector (weight), by the respective column total of the original comparison matrix. Table 4.2 demonstrates the calculation of the maximum eigenvalue also called maximum lambda (λMax). In order to verify that the consistency index (CI) is adequate, Saaty (2000) suggests what is called consistency ratio (CR) which is determined by the ratio between the consistency index and random consistency index (RI). The matrix will be considered consistent if the result of the ratio is less than 10%. The random index value is fixed and is based on the number of evaluated criteria as shown in table 1. In the case of the determinant of customers` churn decision criteria, the consistency rate for the initial group criteria is: CR= �� �� = 0.0320/ 1.32 = 0.0242 < 3% Since its value is less than 10%, the matrix is considered to be consistent. Therefore, considering the eigen vector values / priority weight of determinant of customers churn decision criteria, it is evident that inefficient data/ internet plan criteria have contributed 18.81% to the goal, while mobile number portability implementation criterion contributes (9.73%) to the goal (determinant of customers` churn decision in the Nigeria telecommunication industry). International Journal of Figure 1 – Proposed hierarchical model for customers` churn deci telecommunication Source: AHP model developed by Researchers (2014) U n w an te d ca ll s & T ex ts D is p u te d B il li n g M N P Im p le m en ta ti on F re q u en t P ro m ot io n / b on u se s P oo r In tr a/ In te r C on n ec ti vi ty In ef fi ci en t D at a/ In te rn et P la n P o o r C o m p la in t M a n a g em en t D et er m in a n ts o f cu st om er ch u rn d ec is io n in t h e N ig er ia T el ec o m m u n ic a ti on I n d u st ry International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 Proposed hierarchical model for customers` churn decision in the Nigeria mobile Keys FRIT - Frequently receiving irrelevant texts FRIC - Frequently receiving irrelevant calls SCSC - Disputed Billing short change in service charged. CSNU - Changed for services not used by operator EPMF - Easiness to port with MNP framework CBMI - Competition brought by MNP implementation. CFP - Competitors’ frequent promos BCC - Bonuses on calls/SMS/data plan by competitor DPPO - Dubious promo of present operator DMCSN - Difficulties in making to same network. DMCON - Difficulties in making calls/sending SMS to other network. SDPHC - Sufficient data with high cost. IDPLC - Insufficient data plan with low cost. SANRCC - Service Agent non- customer challenges. SANCC - Service agent non-response to customer complaint. SABE - Service agent being elusive. Source: AHP model developed by Researchers (2014) F R IT F R IC E P M F C B M T D M C S N D M C O N S A N R C C S A B E S A N C C S C S C C S N U C F P D P P O B C C S D P H C ID P L C 87 14, 201 sion in the Nigeria mobile Frequently receiving irrelevant texts. Frequently receiving irrelevant calls. Disputed Billing short change in service Changed for services not used by operator. Easiness to port with MNP framework. Competition brought by MNP Competitors’ frequent promos. Bonuses on calls/SMS/data plan by competitor. Dubious promo of present operator. Difficulties in making calls/sending SMS Difficulties in making calls/sending SMS Sufficient data with high cost. Insufficient data plan with low cost. -willing to resolve response to customer Service agent being elusive. 88 Table 2 – Reduced matrix for determinant of customers churn criteria in the Nigeria Telecommunication Industry Decision Criteria Unwant ed calls & Texts Disput ed Billing MNP Impleme ntation Frequent Promotion / bonuses Poor Intra/In ter Connec tivity Ineffici ent Data/In ternet Plan Poor Compla int Manag ement Weight Unwanted calls & Texts 1.0000 1.1723 1.9663 1.0900 0.5000 0.5391 1.0217 0.1338 λmax 7.1919 Disputed Billing 0.8530 1.0000 1.9593 1.3549 1.1181 0.7115 1.5191 0.1596 CI 0.0320 MNP Implementation 0.5086 0.5104 1.0000 0.9532 0.6151 0.7162 0.6938 0.0973 CR 0.0238 Frequency of Promotion/Bonu ses 0.9174 0.7380 1.0491 1.0000 1.0766 0.8944 1.5675 0.1397 Poor Intra/Inter Connectivity 2.0000 0.8944 1.6258 0.9288 1.0000 1.1277 2.0693 0.1799 Inefficient Data/Internet Plan 1.8549 1.4055 1.3962 1.1181 0.8868 1.0000 2.4151 0.1881 Poor Complaint Management 0.9788 0.6583 1.4414 0.6379 0.4833 0.4141 1.0000 0.1016 Table 3 – Calculation of the maximum eigenvalue of the seven criteria about determinant of customer churn in Nigeria telecommunication industry Criteria Unwanted calls & Texts Dispute billing MNP Implemen tation Frequency of promotion and bonuses Poor Inter/Intra connectivity Inefficient data/intern et plan Poor complaint Management Eigen vector/priorit y weight 0.1338 0.1596 0.0973 0.1397 0.1799 0.1881 0.1016 Column sum 8.1127 6.3789 10.4381 7.0829 5.6799 5.403 10.2865 Maximum eigenvalue/(λMax): (0.1338*8.1127) + (0.1596*6.3789) + (0.0973*10.4381) + (0.1397*7.0829) + (0.1799*5.6799) + (0.1881*5.403) + (0.1016*10.2865) = 7.1919 The test for consistency of churn decision criteria is calculated using the formula below: CI = (λMax-n)/(n-1) CI = (7.1919-7) / 7-1 = 0.1919/6 = 0.0320 International Journal of Table 4 – Reduced matrix for unwanted calls & texts alternatives Unwanted calls and Texts Frequently receiving irrelevant texts Frequently receiving irrelevant texts Frequently receiving irrelevant calls Table 5 – The calculation of the maximum eigenvalue alternatives Decision Alternative of Unwanted calls and Texts Eigen vector/priority weight Column sum Maximum Eigen Value (λmax) In considering the decision alternat priority weight was computed and relation to unwanted calls/texts calls/text, frequently receiving irrelevant text (FRIT) has a weight of 81.23% relative to unwanted calls/text criteria. A positive evaluati (four) times more than a positive evaluation on frequently receiving irrelevant calls alternative (18.77%). Following the procedure of AHP, there is need to check for data inconsistencies. The main objective is to capture enough information to determine whether the customers have been consistent in their choices. The inconsistency index is based on maximum lambda, which is calculated by summing the product of each element in the eigenvector (weight), by the respective column total of the original comparison matrix. Table calculation of the maximum eigenvalue also called maximum lambda ( The test of consistency is done using the formula below CI = (λMax-n)/(n-1) CI = (2.000-2) /2 -1 = 0/1 = 0.0000. In verifying that the consistency index is adequate, Saaty (2000) suggests what is called consistency ratio (CR) which is determined by the ratio between the consistency index and random consistency index (RI). The matrix 10%. The random index value is fixed and is based on the number of evaluated criteria as shown in table 1 in chapter three texts, the consistency rate for the initial group criteria is CR= �� �� = 0.00 Since its value is less than 10% the matrix is considered to be consistent. International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 Reduced matrix for unwanted calls & texts alternatives Frequently receiving irrelevant texts Frequently receiving irrelevant calls 1.0000 4.3282 0.2310 1.0000 The calculation of the maximum eigenvalue for unwanted calls and texts Frequently receiving irrelevant texts Frequently receiving irrelevant calls 0.8123 0.1877 1.2310 5.3282 λmax = { (0.8123*1.2310)+(0.1877*5.3282)}= = 1.0000 +1.0001 = 2.0000 onsidering the decision alternatives of unwanted calls/text criteria, the eigenvector was computed and shows the contribution of each decision alternative relation to unwanted calls/texts criteria. Based on the decision alternatives of frequently receiving irrelevant text (FRIT) has a weight of 81.23% relative to unwanted calls/text criteria. A positive evaluation on this factor contributes approximately 4 (four) times more than a positive evaluation on frequently receiving irrelevant calls alternative (18.77%). Following the procedure of AHP, there is need to check for data inconsistencies. s to capture enough information to determine whether the customers have been consistent in their choices. The inconsistency index is based on maximum lambda, which is calculated by summing the product of each element in the eigenvector (weight), by column total of the original comparison matrix. Table 5 maximum eigenvalue also called maximum lambda (λMax The test of consistency is done using the formula below the consistency index is adequate, Saaty (2000) suggests what is called consistency ratio (CR) which is determined by the ratio between the consistency index and random consistency index (RI). The matrix is considered consistent if the ratio is less than 10%. The random index value is fixed and is based on the number of evaluated criteria as in chapter three. In the case of the decision alternative of unwanted calls and rate for the initial group criteria is Since its value is less than 10% the matrix is considered to be consistent. 89 14, 201 Weight λmax = 2.0000 0.8123 CI = 0.0000 0.1877 CR = 0.0000 for unwanted calls and texts Frequently receiving irrelevant calls 0.1877 5.3282 (0.8123*1.2310)+(0.1877*5.3282)}= criteria, the eigenvector decision alternatives in criteria. Based on the decision alternatives of unwanted frequently receiving irrelevant text (FRIT) has a weight of 81.23% relative to on on this factor contributes approximately 4 (four) times more than a positive evaluation on frequently receiving irrelevant calls alternative (18.77%). Following the procedure of AHP, there is need to check for data inconsistencies. s to capture enough information to determine whether the customers have been consistent in their choices. The inconsistency index is based on maximum lambda, which is calculated by summing the product of each element in the eigenvector (weight), by 5 demonstrates the ax). the consistency index is adequate, Saaty (2000) suggests what is called consistency ratio (CR) which is determined by the ratio between the consistency index and considered consistent if the ratio is less than 10%. The random index value is fixed and is based on the number of evaluated criteria as ernative of unwanted calls and Since its value is less than 10% the matrix is considered to be consistent. 90 Therefore, considering the eigen vector values / priority weight of decision alternatives of unwanted calls/text, it is evident that frequently receiving irrelevant texts alternative have contributed 81.23% to the unwanted calls/text criteria, while frequently receiving irrelevant calls alternative contributes 18.77% in customers decision to churn a network provider. Thus, frequently receiving unsolicited texts messages from network provider can influence customers’ decision to churn more than irrelevant calls. Subscribers are more uncomfortable with frequent irrelevant text messages from services provider. Table 6 – Reduced matrix of dispute billing alternatives Dispute Billing Short-change in service charged Charged for services not used by operator Weight λmax = 2.0000 Short-change in service charged 1.0000 1.1861 0.5426 CI = 0.0000 Charged for services not used by operator 0.8431 1.0000 0.4574 CR = 0.0000 Considering the decision alternatives of dispute billing criteria, the eigenvector / priority, weight was calculated, and shows the contribution of each decision alternatives in relation to dispute billing criteria. Based on the decision alternatives of dispute billing, short change in service charged (SCSC) has a weight of 54.26% on disputed billing criteria. A positive evaluation of this factor contributes approximately one (1) more than a positive evaluation on charge for services not used by operator alternative (45.74%). Following the procedure of AHP, there is need to check for data inconsistencies. Checking for data inconsistencies as before, Table 7 presents the computation of the maximum eigen value (λMax), after which the consistency ratio is calculated as usual. Table 7 – The calculation of the maximum eigenvalue for disputed billing alternatives Disputed Billing criteria Short-change in service charged (SCSC) Charged for services not used by operator Eigen vector/ priority weight 0.5426 0.4574 Total sum 1.8431 2.1861 Maximum Eigenvalue/(λMax) (0.5426*1.8431) + (0.4574*2.1861) = 2.0000 The test of consistency is calculated using the formula below: CI = (λMax-n)/(n-1) CI = (2.000-2) /2 -1 = 0/1 = 0.0000 As shown in earlier computations, the consistency ratio of disputed billing criteria, using the initial group criteria is: CR= �� �� = 0.00 Since the CR value is less than 10%, the matrix is considered to be consistent. Therefore, in considering the eigen vector values / priority weight of decision alternatives of disputed billing, it is evident that short-change in service charged (SCSC) alternative have International Journal of contributed 54.26% to the disputed billing criteria, while charg used by operator alternative, contributes 45.74% to disputed billing criterion. Table 8 – Reduced matrix for mobile number portability alternatives Mobile Number Portability (MNP) Easiness to port Easiness to port with MNP framework Competition brought by MNP implementation Considering the decision alternat priority weight was calculated relation to MNP implementation portability implementation, easiness to port with MNP framework with regards to MNP implementation criteria. A positive evaluation of this factor contributes approximately twice more than a positive evaluation of c implementation alternative, which has a weight of 37.68%. Following the procedure of AHP, there is need to check for data inconsistencies. As has been done in previous analyses, the inconsistency index is computed as shown in Table Table 9 – The calculation of the maximum eigenvalue for MNP alternatives Mobile Number Portability Implementation criteria Eigen vector/ priority weight Total sum Maximum Eigenvalue/(λMax) The test of consistency is calculated by using the formula below CI = (λMax-n)/(n-1) CI = (2.000-2) /2 -1 = 0/1 = 0.0000 Thus, in the case of the mobile number portability implementation criteria, the consistency rate for the initial group criteria is CR= �� �� = 0.00 CR which is less than 10%, indicates that the matrix is consistent. Therefore, in considering eigen mobile number portability implementation, it is evident that easiness to port with MNP framework alternative have contributed 62.32% to the mobile number portability International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 contributed 54.26% to the disputed billing criteria, while charging customers alternative, contributes 45.74% to disputed billing criterion. Reduced matrix for mobile number portability alternatives Easiness to port with MNP framework Competition brought by MNP implementation Weight 1.0000 1.6542 0.6232 0.6045 1.0000 0.3768 Considering the decision alternatives of MNP implementation criteria, the eigenvector was calculated, and shows the contribution of each decision alternative MNP implementation criteria. Based on the decision alternatives of asiness to port with MNP framework has a weight of 62.32% with regards to MNP implementation criteria. A positive evaluation of this factor contributes approximately twice more than a positive evaluation of competition brought by MNP alternative, which has a weight of 37.68%. Following the procedure of AHP, there is need to check for data inconsistencies. As has been done in previous analyses, the inconsistency index is computed as shown in Table 9. The calculation of the maximum eigenvalue for MNP alternatives Easiness to port with MNP framework Competition brought by MNP implementation 0.6232 1.6045 (0.6232*1.6045) + (0.3768*2.6542) = The test of consistency is calculated by using the formula below: Thus, in the case of the mobile number portability implementation criteria, the consistency rate for the initial group criteria is CR which is less than 10%, indicates that the matrix is consistent. Therefore, in considering eigen vector values / priority weight of decision alternatives of mobile number portability implementation, it is evident that easiness to port with MNP framework alternative have contributed 62.32% to the mobile number portability 91 14, 201 ing customers for services not alternative, contributes 45.74% to disputed billing criterion. Weight λmax = 2.0000 0.6232 CI = 0.0000 0.3768 CR = 0.0000 criteria, the eigenvector / decision alternatives in criteria. Based on the decision alternatives of mobile number has a weight of 62.32% with regards to MNP implementation criteria. A positive evaluation of this factor contributes tition brought by MNP alternative, which has a weight of 37.68%. Following the procedure of AHP, there is need to check for data inconsistencies. As has been done in previous analyses, the The calculation of the maximum eigenvalue for MNP alternatives Competition brought by MNP implementation 0.3768 2.6542 = 2.0000 Thus, in the case of the mobile number portability implementation criteria, the consistency vector values / priority weight of decision alternatives of mobile number portability implementation, it is evident that easiness to port with MNP framework alternative have contributed 62.32% to the mobile number portability 92 implementation criteria. However, competition brought by MNP implementation alternative contributes 37.68% to the mobile number portability implementation criterion. Table 10 – Reduced matrix for frequent promotions/ bonuses alternatives Frequent Promotion/ bonuses Competitor’ frequent promos Bonuses on calls/SMS/data plan by competitor Dubious promo of present operator Weight Competitors’ frequent promos 1.0000 1.1397 1.0582 0.3518 λmax = 3.0347 Bonuses on calls/SMS/data plan by competitor 0.8774 1.0000 1.6149 0.3710 CI = 0.0174 Dubious promo of present operator 0.9450 0.6192 1.0000 0.2772 CR = 0.0299 Considering the decision alternatives of frequent promotions/ bonuses criteria, the eigenvector priority weight was calculated and shows the contribution of each of the decision alternatives in relation to frequent promotions/ bonuses criteria. Based on the decision alternatives of frequent promotions/ bonuses, bonuses on calls/SMS/data plan by competitor has a weight of 37.10% of frequent promotions /bonuses criteria. A positive evaluation of these factors contributes approximately twice more than a positive evaluation of dubious promo of present operator alternative (27.72%). Table 11 – The calculation of the maximum eigenvalue for frequent promotions/ bonuses by competitors’ alternatives Frequent promotion /bonuses Competitors’ frequent promos Bonuses on calls/SMS/data plan by competitor Dubious promo of present operator Eigenvector/ priority weight 0.3518 0.3710 0.2772 Total sum 2.8224 2.7589 3.6731 Maximum Eigenvalue/(λMax) (0.3518*2.8224) + (0.3710*2.7589) + (0.2772*3.6731) = 3.0347 The test of consistency was calculated as follows: CI = (λMax-n)/(n-1) CI = (3.0347-3) / 3 - 1 = 0.0347 / 2 = 0.0174 In the case of the frequent promotion/bonuses criteria, the consistency rate for the initial group criteria is CR= �� �� = 0.0174/ 0.58 = 0.0299 < 3% International Journal of Since CR value is less than 10%, the matrix is consistent. Therefore, in considering the eigen val frequent promotion/bonuses, it is evident that b alternative have contributed 37.10% to the frequent promotion/bonuses criteria, while dubious promo of present operator alternative contributes 27.72% to criterion. Table 12 – Reduced matrix of poor inter/intra connectivity alternatives Poor Intra/Inter Connectivity Difficulties in making calls/sending SMS to same network Difficulties in making calls/sending SMS to same network Difficulties in making calls/sending SMS to other network Considering the decision alternat eigenvector priority weight was computed and alternatives in relation to poor inter/intra connectivity alternatives of poor inter/intra connectivity network have a weight of 73.92% on poor inter/intra connectivity criteria. A positive evaluation of this factor contributes approximately thrice more than a positive evaluation of difficulties in making calls/sending SMS to other Table 13 – The calculation of the maximum eigenvalue for poor inter/intra connectivity alternatives Poor inter/intra connectivity criteria Eigenvector/ priority weight Total sum Maximum Eigenvalue/(λMax) (0.7392*1.3529)+(0.2608*3.8338) = 2.0000 The test of consistency was calculated as follows: CI = (λMax-n)/(n-1) CI = (2.000 - 2) /2 -1 = 0/1 = 0.0000 In the case of the poor inter/intra connectivity criteria, the consistency rate for the initial group criteria is: International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 Since CR value is less than 10%, the matrix is consistent. Therefore, in considering the eigen values / priority weight of decision alternatives of frequent promotion/bonuses, it is evident that bonuses on calls/SMS/data plan by competitor alternative have contributed 37.10% to the frequent promotion/bonuses criteria, while dubious alternative contributes 27.72% to frequent promotion/bonuses Reduced matrix of poor inter/intra connectivity alternatives Difficulties in making calls/sending SMS to same network Difficulties in making calls/sending SMS to other network Weight 1.0000 2.8338 0.7392 0.3529 1.0000 0.2608 Considering the decision alternatives of poor inter/intra connectivity was computed and shows the contribution of each of s in relation to poor inter/intra connectivity criteria. Based on th poor inter/intra connectivity, difficulties in making calls/sending SMS to same have a weight of 73.92% on poor inter/intra connectivity criteria. A positive evaluation of this factor contributes approximately thrice more than a positive evaluation of making calls/sending SMS to other network alternative (26.08%). The calculation of the maximum eigenvalue for poor inter/intra connectivity Difficulties in making calls/sending SMS to same network Difficulties in making calls/sending to other net 0.7392 1.3529 (0.7392*1.3529)+(0.2608*3.8338) = 2.0000 The test of consistency was calculated as follows: In the case of the poor inter/intra connectivity criteria, the consistency rate for the initial 93 14, 201 ues / priority weight of decision alternatives of onuses on calls/SMS/data plan by competitor alternative have contributed 37.10% to the frequent promotion/bonuses criteria, while dubious frequent promotion/bonuses Weight λmax = 2.0000 0.7392 CI = 0.0000 0.2608 CR = 0.0000 ives of poor inter/intra connectivity criteria, the shows the contribution of each of the decision criteria. Based on the decision ifficulties in making calls/sending SMS to same have a weight of 73.92% on poor inter/intra connectivity criteria. A positive evaluation of this factor contributes approximately thrice more than a positive evaluation of alternative (26.08%). The calculation of the maximum eigenvalue for poor inter/intra connectivity Difficulties in making calls/sending SMS to other network 0.2608 3.8338 In the case of the poor inter/intra connectivity criteria, the consistency rate for the initial 94 CR= �� �� = 0.00 Since CR value is less than 10%, the matrix is considered to be consistent. Thus, in considering the eigen values/priority weight of decision alternatives of poor inter/ intra connectivity, it is evident that difficulties in making calls/sending SMS to same network alternative have contributed 73.92% to the mobile number portability implementation criterion, while, difficulties in making calls/sending SMS to other network alternative contribute 26.08% to the poor inter/intra connectivity criterion. Table 14 – Reduced matrix of inefficient data/ internet plan criteria Inefficient Data/Internet Plan Sufficient data with high cost Insufficient data plan with low cost Weight λmax = 2.0000 Sufficient data with high cost 1.0000 2.7231 0.7314 CI = 0.0000 Insufficient data plan with low cost 0.3672 1.0000 0.2686 CR = 0.0000 Considering the decision alternatives of inefficient data/ internet plan criteria, the eigenvector priority weight was calculated and shows the contribution of each decision alternative in relation to inefficient data/ internet plan criteria. Based on the decision alternatives of inefficient data/ internet plan, sufficient data with high cost have a weight of 73.14% of inefficient data/internet plan criteria. A positive evaluation of this factor contributes approximately thrice (3 times) more than a positive evaluation of insufficient data plan with low cost alternative (26.86%). Table 15 – The calculation of the maximum eigenvalue for inefficient data/ internet plan alternatives Inefficient data/internet plan Sufficient data with high cost Insufficient data plan with low cost Eigen vector/ priority weight 0.7314 0.2686 Total sum 1.3672 3.7231 Maximum Eigenvalue/(λMax) (0.7314*1.3672) + (0.2686*3.7231) = 2.0000 The test of consistency is done using the formula below: CI = (λMax-n)/(n-1) CI = (2.000-2) /2 -1 = 0/1 = 0.0000 In the case of the inefficient data/internet plan criteria, the consistency rate for the initial group criteria is: CR= �� �� = 0.00 International Journal of Since the value of the CR is less than 10%, the matrix is considered to be consistent. Therefore, in considering the eigen values / priority weight of decision alternatives of inefficient data/internet plan, it is evident that sufficient data plan with high cost alternative have contributed 73.14% to the inefficient data/ internet plan criteria, while inefficient plan with low cost alternative contributes 26.86% to Table 16 – Reduced matrix of Poor Complaint Management Poor Complaint Management Service Agent non resolve customer challenges Service Agent non- willing to resolve customer challenges Service agent non- response to customer complaint Service agent being elusive Considering the decision alternat eigenvector priority weight was calculated and alternative in relation to poor complaint management alternatives of poor complaint management, customer challenges have a weight of 51.46% with regards to poor complaint m criteria. A positive evaluation on this factor contributes approximately thrice more than a positive evaluation of service agent being elusive Table 17 – The calculation of the maximum eigenvalue alternatives Poor complaint management Service Agent non customer challenges Eigen vector/ priority weight Total sum Maximum Eigenvalue/(λMax) The test of consistency is calculated below: CI = (λMax-n)/(n-1) CI = (3.0394-3) /3 -1 = 0.0394/2 = 0.0197 In the case of the poor complaint management criteria, the consistency rate for the initial group criteria is International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 Since the value of the CR is less than 10%, the matrix is considered to be consistent. ng the eigen values / priority weight of decision alternatives of inefficient data/internet plan, it is evident that sufficient data plan with high cost alternative have contributed 73.14% to the inefficient data/ internet plan criteria, while inefficient plan with low cost alternative contributes 26.86% to the inefficient data/internet plan criteria. Reduced matrix of Poor Complaint Management Service Agent non-willing to resolve customer challenges Service agent non-response to customer complaint Service agent being elusive 1.0000 2.2423 2.0880 0.4460 1.0000 1.6229 0.4789 0.6162 1.0000 Considering the decision alternatives of poor complaint management was calculated and shows the contribution of each in relation to poor complaint management criteria. Based on th poor complaint management, service agent non-willing have a weight of 51.46% with regards to poor complaint m criteria. A positive evaluation on this factor contributes approximately thrice more than a ervice agent being elusive alternative (20.69%). The calculation of the maximum eigenvalue for poor complaint management Service Agent non-willing to resolve customer challenges Service agent non-response to customer complaint 0.5146 0.2784 1.9249 3.8585 (0.5146*1.9249) + (0.2784*3.8585) + (0.2069*4.7109) = 3.0394 The test of consistency is calculated below: In the case of the poor complaint management criteria, the consistency rate for the initial 95 14, 201 Since the value of the CR is less than 10%, the matrix is considered to be consistent. ng the eigen values / priority weight of decision alternatives of inefficient data/internet plan, it is evident that sufficient data plan with high cost alternative have contributed 73.14% to the inefficient data/ internet plan criteria, while inefficient data the inefficient data/internet plan criteria. Weight 0.5146 λmax = 3.0394 0.2784 CI = 0.0197 0.2069 CR = 0.0339 ives of poor complaint management criteria, the hows the contribution of each decision criteria. Based on this decision willingness to resolve have a weight of 51.46% with regards to poor complaint management criteria. A positive evaluation on this factor contributes approximately thrice more than a for poor complaint management Service agent being elusive 0.2069 4.7109 (0.5146*1.9249) + (0.2784*3.8585) + (0.2069*4.7109) = 3.0394 In the case of the poor complaint management criteria, the consistency rate for the initial 96 CR= �� �� = 0.0197/ 0.58 = 0.0339 < 4% Again since the CR value is less than 10%, the matrix is considered to be consistent. Therefore, in considering eigen vector values / priority weight of decision alternatives of poor complaint management, it is evident that service agent non-willingness to resolve customer challenges alternative contributed 51.46% to the poor complaint management criteria, while service agent being elusive alternative contributes 20.69% to poor complaint management criteria. Table 18 shows the priorities of the criteria with respect to the main goal which is to determine reasons for customers leaving a particular network provider to another network provider in the Nigeria telecommunication industry. Based on the views of the customers, the most critical determinant factor for leaving a network provider to another network provider is inefficient data/internet plan with priority 0.1881, followed by poor inter/intra connectivity with priority 0.1799, dispute in billing with priority 0.1569, frequent promotion/ bonuses with priority 0.1397, unwanted calls/ texts with priority 0.1338, and poor complaint management with priority 0.1016, while the least determinant factor is the MNP implementation with priority 0.0973. Table 18 – Composite priorities of the criteria about Goal The horizontal bar chart in figure 2 represents the pictorial diagram of decision criteria where the horizontal bar length is the priority of each criterion. From the chart, inefficient data plan/ internet plan has the longest bar with priority of 0.1881, followed by poor inter/ intra connectivity with priority of 0.1799, dispute billing with priority of 0.1596, frequent promotion and bonuses with priority of 0.1397, unwanted calls and texts with priority of 0.1338 and poor complaints management with priority of 0.1016 while the MNP implementation has the shortest bar with priority of 0.0973. Goal: determi- nants of cus- tomer churn in Nigeria Tele- communication industry Un- wanted calls/ texts Dispute billing MNP Im- plementa- tion Frequent promo- tion/ bo- nuses Poor in- ter/intra connec- tivity Ineffi- cient data/ internet plan Poor com- plaint manage- ment Pooled Average Composite Priority 0.1338 0.1569 0.0973 0.1397 0.1799 0.1881 0.1016 Relative Preference Ranking 5th 3rd 7th 4th 2nd 1st 6th International Journal of Figure 2 – Bar chart showing decision criteria with their corresponding priority Table 19 reveals the customers perception with regards to the decision alternatives of unwanted calls/texts using the composite priorities. This criterion has only two alternatives in this study. Customers` mostly 0.8123 than frequently receiving irrelevant calls (FRIC) with priority of 0.1877. Table 19 – Composite priorities of the decision alternative about unwanted calls/ texts Decision alternative with regards to unwanted calls/ texts Pooled Average Composite priority Relative preference ranking The horizontal bar chart in figure alternatives of unwanted calls/text where the horizontal bar length is the priority of each criterion. From the chart, frequently receiving irrelevant texts has the longest bar with priority of 0.8123, compared to frequen Figure 3 – Bar chart showing decision alternatives of unwanted calls/texts Table 20 reveals subscribers` using the composite priorities. service charge with priority operators with priority of 0.4574 International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 Bar chart showing decision criteria with their corresponding priority reveals the customers perception with regards to the decision alternatives of unwanted calls/texts using the composite priorities. This criterion has only two alternatives in mostly detest frequently receiving irrelevant texts (FRIT) with priority frequently receiving irrelevant calls (FRIC) with priority of 0.1877. Composite priorities of the decision alternative about unwanted calls/ texts Frequently receiving irrelevant text (FRIT) Frequently receiving irrelevant calls (FRIC) 0.8123 1 The horizontal bar chart in figure 3 represents the pictorial diagram of decision alternatives of unwanted calls/text where the horizontal bar length is the priority of each frequently receiving irrelevant texts has the longest bar with priority compared to frequently receiving irrelevant calls with priority of Bar chart showing decision alternatives of unwanted calls/texts subscribers` perception about the decision alternatives of dispute billing ties. Subscribers mostly preferred disputed billing short change in service charge with priority of 0.5426 compared to charged for services not used by the operators with priority of 0.4574. 97 14, 201 Bar chart showing decision criteria with their corresponding priority. reveals the customers perception with regards to the decision alternatives of unwanted calls/texts using the composite priorities. This criterion has only two alternatives in xts (FRIT) with priority frequently receiving irrelevant calls (FRIC) with priority of 0.1877. Composite priorities of the decision alternative about unwanted calls/ texts Frequently receiving irrelevant calls (FRIC) 0.1877 2 the pictorial diagram of decision alternatives of unwanted calls/text where the horizontal bar length is the priority of each frequently receiving irrelevant texts has the longest bar with priority of 0.1877. perception about the decision alternatives of dispute billing preferred disputed billing short change in 0.5426 compared to charged for services not used by the 98 Table 20 – Composite priorities of the decision alternative about dispute in billing Decision alternative with regards to dispute billing Disputed billing short change on service charge Charged for services not used by operator Pooled Average Composite priority 0.5426 0.4574 Relative preference ranking 1 2 The horizontal bar chart in figure 4 represent the pictorial diagram of decision alternatives of dispute billing where the horizontal bar length is the priority of each criterion. From the bar chart, one observes that disputed billing in short change of service charged has the longest bar with priority 0.5426 compared to charge for services not used by operator with priority 0.4574. Figure 4 – Bar chart showing decision alternatives of disputed billing Table 21 reveals the customers perception about the decision alternatives of MNP implementation using the composite priorities. The customers mostly preferred easiness to port with MNP framework, with a priority of 0.6232, compared to competition brought by MNP framework with priority of 0.3768. Table 21 – Composite priorities of the decision alternative about MNP Implementation Decision alternative with regards to MNP implementation Easiness to port with MNP framework Competition brought by MNP implementation Pooled Average Composite priority 0.6232 0.3768 Relative preference ranking 1 2 The horizontal bar chart in figure 5 represents the pictorial diagram of decision alternatives of MNP implementation where the horizontal bar length is the priority of each criterion. From the chart, easiness to port using MNP framework has the longest bar with priority of 0.6232, compared to competition brought by MNP implementation with priority of 0.3768. International Journal of Figure 5 – Bar chart showing decision alternatives of MNP implementation Table 22 reveals the customers perception about the decision alternatives of frequent promotions/ bonuses using the co study. The customers mostly priority of 0.3710, followed by competitor’s frequent promos with priority least preferred is dubious promo of present operator with priority of 0.2772. Table 22 – Composite priorities of the decision alternative about frequent promotions/ bonuses Decision alternative with regards to frequent promotions/ bonuses Pooled Average Composite priority Relative preference ranking The horizontal bar chart alternatives of frequent promotions and bonuses where the horizontal bar length is the priority of each criterion. As the chart the longest bar with priority priority of 0.3518, while dubious promotion by present operators has the shortest bar length with priority of 0.2772. Table 23 reveals the customers perception about the decision alternatives of poor intra connectivity using the composite priorities. The customers difficulty in making calls/sending SMS to same network with priority difficulties in making calls/SMS to other network with priority of 0 The horizontal bar chart in figure of poor inter/intra connectivity where the horizontal bar length is the priority of each criterion. As the chart shows, one observes network has the longest bar with priority calls/sending SMS to other network with priority International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 Bar chart showing decision alternatives of MNP implementation reveals the customers perception about the decision alternatives of frequent promotions/ bonuses using the composite priorities. This criterion has three alternatives in this preferred bonuses on calls/SMS/data plan by competitors with 0.3710, followed by competitor’s frequent promos with priority eferred is dubious promo of present operator with priority of 0.2772. Composite priorities of the decision alternative about frequent promotions/ Competitors’ frequent promos Bonuses on calls/SMS/ data plan by competitors 0.3518 0.3710 2 1 bar chart in figure 6 represents the pictorial diagram of decision alternatives of frequent promotions and bonuses where the horizontal bar length is the priority chart indicates, bonuses on calls/SMS/data plan by competitors has of 0.3710 followed by competitors` frequent promotion with dubious promotion by present operators has the shortest bar length reveals the customers perception about the decision alternatives of poor intra connectivity using the composite priorities. The customers are most difficulty in making calls/sending SMS to same network with priority of 0.7392 compared to difficulties in making calls/SMS to other network with priority of 0.2608. chart in figure 7 stand for the pictorial diagram of decision alternatives of poor inter/intra connectivity where the horizontal bar length is the priority of each criterion. shows, one observes that difficulties in making calls/sending SMS to the same network has the longest bar with priority of 0.7392, compared to difficulties in making calls/sending SMS to other network with priority of 0.2608. 99 14, 201 Bar chart showing decision alternatives of MNP implementation reveals the customers perception about the decision alternatives of frequent mposite priorities. This criterion has three alternatives in this preferred bonuses on calls/SMS/data plan by competitors with 0.3710, followed by competitor’s frequent promos with priority of 0.3518 and the eferred is dubious promo of present operator with priority of 0.2772. Composite priorities of the decision alternative about frequent promotions/ Dubious promo of present operator 0.2772 3 the pictorial diagram of decision alternatives of frequent promotions and bonuses where the horizontal bar length is the priority bonuses on calls/SMS/data plan by competitors has frequent promotion with dubious promotion by present operators has the shortest bar length reveals the customers perception about the decision alternatives of poor inter/ mostly disturbed by 0.7392 compared to the pictorial diagram of decision alternatives of poor inter/intra connectivity where the horizontal bar length is the priority of each criterion. aking calls/sending SMS to the same compared to difficulties in making 100 Figure 6 – Bar chart showing decision `alternative of frequent promotions and bonuses Table 23 – Composite priorities of the decision alternative about poor inter/intra connectivity Decision alternative with regards to poor inter/ intra connectivity Difficulties in making calls / sending SMS to same network Difficulties in making calls/sending SMS to other network Pooled Average Composite priority 0.7392 0.2608 Relative preference ranking 1 2 Figure 7 – Bar chart showing decision alternatives of poor inter/intra connectivity Table 24 reveals the customers` perception about the decision alternatives of inefficient data/ internet plan using the composite priorities. The customers mostly preferred sufficient data with high cost with priority of 0.7314, compared to insufficient data with low cost priority of 0.2686. The horizontal bar chart in figure 8 represents the pictorial diagram of decision alternatives of inefficient data/ internet plan where the horizontal bar length is the priority of each criterion. From the chart, sufficient data plan with high cost has the longest bar with priority of 0.7314 compared to insufficient data plan with low cost priority of 0.2686. International Journal of Table 24 – Composite priorities of the decision alternative about inefficient data/internet plan Decision alternative with regards to inefficient data /internet plan Pooled Average Composite priority Relative preference ranking Figure 8 – Bar chart showing decision alternatives of inefficient data/ internet plan Table 25 shows the customers perception about the decision alternatives of poor complaint management using the composite priorities. alternatives. The customers most preferred service agent not willing to resolve customer challenges with priority of 0.5146 complaint with priority of 0.2784 priority of 0.2069. Table 25 – Composite priorities of the decision alternative about poor complaint management Decision alternative with regards to poor complaint management Pooled Average Composite priority Relative preference ranking The horizontal bar chart in f alternatives of poor complaint management where the horizontal bar length is the priority of each criterion. From the chart customer challenges has the longest b non- response to customers` elusive has the shortest bar length with International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 Composite priorities of the decision alternative about inefficient data/internet plan with regards to inefficient data /internet plan Insufficient data with low cost Sufficient data with Pooled Average Composite priority 0.2686 2 Bar chart showing decision alternatives of inefficient data/ internet plan the customers perception about the decision alternatives of poor complaint management using the composite priorities. In this study, this criterion has three ternatives. The customers most preferred service agent not willing to resolve customer 0.5146, followed by service agent non-res 0.2784, while the least preferred is service agent bein Composite priorities of the decision alternative about poor complaint management Service Agent non-willing to resolve customer challenges Service agent non response to customer complaint Pooled Average Composite priority 0.5146 0.2784 1 2 chart in figure 9 represents the pictorial diagram of decision alternatives of poor complaint management where the horizontal bar length is the priority of each criterion. From the chart, one observes that service agent non-willing customer challenges has the longest bar with a priority of 0.5146, followed by complaint with a priority of 0.2784 and service agent being elusive has the shortest bar length with a priority of 0.2069. 101 14, 201 Composite priorities of the decision alternative about inefficient data/internet plan Sufficient data with high cost 0.7314 1 Bar chart showing decision alternatives of inefficient data/ internet plan the customers perception about the decision alternatives of poor his criterion has three ternatives. The customers most preferred service agent not willing to resolve customer response to customer the least preferred is service agent being elusive with Composite priorities of the decision alternative about poor complaint management agent non- response to customer Service agent being elusive 0.2069 3 the pictorial diagram of decision alternatives of poor complaint management where the horizontal bar length is the priority of willingness to resolve 0.5146, followed by service agent d service agent being 102 Figure 9 – Bar chart showing decision alternatives of poor complaint management 5. Conclusion The study concluded that AHP is appropriate at modeling complex problem such as customer churn motivations in the service industry (such as telecommunication) by assigning priority to churn drivers in the business environment. The criteria and the alternatives in the AHP model are drivers of customers` churn decision in the Nigeria telecommunication industry. It also provided weight to each criteria as well as the alternatives as they motivates or influences an average subscriber in the study area to leave/abandon a network provider for competitor. The study also shows that MNP was not on the top priority list of factors that make customers to churn network in Nigeria, thereby suggesting that the factor (MNP) is less than effective regulatory policy for stimulating appropriate competition that can yield improved customer service delivery in the Nigeria telecommunication industry. The study also concluded that certain promotional activities by telecommunication services providers are dubious and not complementary to attracting customers to their network but further enhances churning decision of subscriber, which may not be in the best interest of private firms whose primary motives to make profit. 6. Recommendations and Suggestions for future research The following recommendations were drawn from the findings of this study: (i) That GSM operator should improve the quality of their services through appropriate mix of churn drivers in order to increase their market share significantly. (ii) Network operators should engage in the use Analytic hierarchy process model that can better unravel the motivations for customers churn through pairwise comparison of churn drivers by subscribers who have experienced the service for better decision rather than predicting churn from database. (iii) Mobile telecommunication services provider should reverse the churn alternatives with high priority in order to sustain improved market share. (iv) Service delivery of network operators in Nigeria should be geared towards reversing and correcting the criteria and alternatives that motivates customer(s) to churn network provider. (v) Marketing effort should focus more on effective service recovery on criteria and the alternatives motivating subscribers churn in the Nigeria telecommunication industry. (vi) Mobile network telecommunication service providers in Nigeria should avoid dubious promotional means as it trigger subscribers` churn decision, damage organisational reputation and long-term survival. International Journal of This study highlighted a number issues on which further order to extend the frontiers of knowledge. (i) It will be necessary to collect data in all the geo federation to ascertain whether the preferences (weight) for the drivers of customers` churn decision will change significantly in order to overhaul the corporate strategies of the telecommunication firms. (ii) The present research study was limited to GSM sub Telecommunication industry, further research may Access (CDMA) sub-sector, which has subscribers. (iii) The present study has discussed three levels AHP model, which determinant of customers` churn decision in the do well if it provides elaborate analysis of th comparative analysis of the performance of the GSM operators to funded by GSM service providers in Nigeria References 1. Arora, M. (2013). Role of Service Quality in Customer Relationship Management: An Empirical Study of Indian Telecom Industry, Marketing Management Research and Development 2. Burez, J., & Van den Poel, D. (2007 models to reduce customer attrition by targeted marketing for subscript Expert Systems with Applications, 3. Crawford, G.B. & Williams, C. (1985). A note on the analysis of subjective judgment matrices. The geometric mean procedure for estimating the scale of a judgment matrix. Journal of Mathematical Psychology 4. Golden, B. L., Wasli, E, A. & Harker, P, T. (1989). Process:Applications and Studies 5. Kim, H. S., & Yoon, C. H. (2004). Determinants of subscriber churn and customer loyalty in the Korean mobile telephony market. 765. 6. Kolajo, T & Adeyemo, Telecommunications Industry Customer Churn Using both Descriptive and Predictive Algorithms. Computing Inform 27-34. 7. Leary, M. R. (2001). Introduction to behavioural research methods. Heighs. 8. Lu, M. H., Madu, C. N., Kuei, C., & Winokur, D. (1994). Integrating QFD, AHP and benchmarking in strategic marketing. 50. 9. NCC (2014). Subscriber Data, 2001 November, 2014. 10. OECD (2003/2004). African Economic Outlook. Cooperation and Development. International Journal of Economic Behavior, 5, n. 1, pp. 3-14, 201 This study highlighted a number issues on which further research can be carried out in order to extend the frontiers of knowledge. For future studies: It will be necessary to collect data in all the geo-political zones and whether the preferences (weight) for the drivers of customers` churn decision will change significantly in order to overhaul the corporate strategies of the telecommunication firms. The present research study was limited to GSM sub-sector of the Nigerian ndustry, further research may include Code sector, which has also been severely neglected by most The present study has discussed three levels AHP model, which aid evaluation of of customers` churn decision in the telecom industry. Future research elaborate analysis of the forth level of the hierarchy, comparative analysis of the performance of the GSM operators to funded by GSM service providers in Nigeria. Role of Service Quality in Customer Relationship Management: An Empirical Study of Indian Telecom Industry, International Journal of Sales and Marketing Management Research and Development (IJSMMRD), 3(2), 87 ez, J., & Van den Poel, D. (2007). CRM at a Pay-TV Company: Using analytical models to reduce customer attrition by targeted marketing for subscript Expert Systems with Applications, 32(2), 277–288. Crawford, G.B. & Williams, C. (1985). A note on the analysis of subjective judgment matrices. The geometric mean procedure for estimating the scale of a judgment matrix. cal Psychology, 29, 387–405. Golden, B. L., Wasli, E, A. & Harker, P, T. (1989). The Analytic Hierarchy Process:Applications and Studies. New York: Springer Verlag. Kim, H. S., & Yoon, C. H. (2004). Determinants of subscriber churn and customer the Korean mobile telephony market. Telecommunications Policy Kolajo, T & Adeyemo, A. B. 2012). Data Mining Technique for Predicting Industry Customer Churn Using both Descriptive and Predictive Computing Information Systems & Development Informatics Journal. Introduction to behavioural research methods. Lu, M. H., Madu, C. N., Kuei, C., & Winokur, D. (1994). Integrating QFD, AHP and tegic marketing. Journal Business & Industrial Marketing NCC (2014). Subscriber Data, 2001-2014. Accessed: www.ncc.com/subscriberdata.3 African Economic Outlook. France: Organisation for Economic Cooperation and Development. 103 14, 201 research can be carried out in zones and states of the whether the preferences (weight) for the drivers of customers` churn decision will change significantly in order to overhaul the corporate strategies of sector of the Nigerian include Code-Division Multiple also been severely neglected by most aid evaluation of the industry. Future research may e forth level of the hierarchy, to allow for drivers of churn, if Role of Service Quality in Customer Relationship Management: An International Journal of Sales and (IJSMMRD), 3(2), 87-94. TV Company: Using analytical models to reduce customer attrition by targeted marketing for subscription services. Crawford, G.B. & Williams, C. (1985). A note on the analysis of subjective judgment matrices. The geometric mean procedure for estimating the scale of a judgment matrix. The Analytic Hierarchy Kim, H. S., & Yoon, C. H. (2004). Determinants of subscriber churn and customer Telecommunications Policy, 28, 751– 2012). Data Mining Technique for Predicting Industry Customer Churn Using both Descriptive and Predictive ation Systems & Development Informatics Journal. 3(2). Maryland, Neeham Lu, M. H., Madu, C. N., Kuei, C., & Winokur, D. (1994). Integrating QFD, AHP and Journal Business & Industrial Marketing, 9(1), 41 - www.ncc.com/subscriberdata.3, France: Organisation for Economic 104 11. Oyatoye E.O, Adebiyi S.O & Amole, B. B. (2015). Evaluating Subscribers Preference for Service Attributes of Mobile Telecommunication in Nigeria Using Analytical Hierarchy Process (AHP), International Journal of Analytical Hierarchy Process (IJAHP), 7(2), 171-187. 12. Pyramid Research (2010). The Impact of Mobile Services in Nigeria: How Mobile Technologies Are Transforming Economic and Social Activities; Abuja, Nigeria. 13. Reynolds, T. & Jolly, J. (1980). Measuring personal values: an evaluation of alternativemethods. Journal Marketing Science, 17, 531–536. 14. Saaty T. L & Kearns K. P (1991). Analytical planning: The organization of systems, Theanalytic hierarchy process series, RWS Publications, Pittsburgh, USA. 15. Saaty, T. L. (1977). A scaling method for priorities in hierarchical structures. Journal of Mathematical Psychology, 15, 59 – 62. 16. Saaty, T. L. (1980). The Analytic Hierarchy Process, McGraw- Hill, New York. 17. Saaty, T. L. (2000). Fundamentals of Decision Making and Priority Theory with Analytical Hierarchy Process. Pittsburgh, PA: RWS Publications. 18. Sato, Y. (2007). Administrative Evaluation and Public Sector Reform: An Analytic Hierarchy Process Approach, International Transactions in Operational Research, 14(5), 445 - 453. 19. Stephen, A. (1984). Contract management handbook for commercial construction, CA, Naris Publications, 254 - 279. 20. Taylor III, B.W. (2002). Management Science, (7th ed). New Jersey: Pearson Prentice- Hall, Inc. 21. Vaidya, O.S. & Kumar, S. (2006). Analytic Hierarchy Process: Overview of Applications. European Journal of Operational research, 169(1), 1–29. 22. Van-Den-Poel, D. & Lariviere, B. (2004). Customer attrition analysis for financial services using proportional hazard models. European Journal of Operational Research, 157, 196-217. 23. Vargas, L. G. (1990). An overview of the analytic hierarchy process and its applications, Europe Journal of Operation Research, 48(1), 2-8.