Consequences of Customer Engagement in Social Networking Sites : Employing Fuzzy Delphi Technique for Validation

The consequences of the customer engagement in the Social Networking Sites (SNS) community have direct impact on the brand. This present research was conducted to examine the cohesive mechanisms for item verification on the most influential consequences of participating the brand community and joining the electronic Word-of-Mouth (eWOM) as the manifestation of the behavior of such communities. Using Fuzzy Delphi techniques, a total of 12 heterogeneous experts are involved in the verification process through a 7-point linguistic scale of the questionnaire survey. The results show good evidence of expert consensus by reaching 75% for each consequence of the engaged customers. On the SNS platform, further aspects of the inspected effects can be expanded to be studied on relevant domains. Practitioners will be more strategic in maintaining and fostering customer relationships, and consistently influencing new customers when interacting actively through SNS brand pages. Keywords—Customer engagement; fuzzy delphi; SNS; consequences; brand page


INTRODUCTION
The explosive growth of Social Networking Sites (SNS) is the basis for firms to open space for customer engagement in the proliferation of brand-related ideas, information, and thought surrounding their network venture.In the circle of electronic Word-of-Mouth (eWOM) activities, customers and prospects use SNS not just to acquire information but to engage with brand and products [1]- [2].Customers are equipped with the brand pages that are created by firms for marketing and communication activities related to brand and products.These pages are able to tie existing customers and catch the prospect's attention due to the firms' series of repeated and effective strategies, for an instance, posting the good copywriting to advertise a newly-launched product.
Notwithstanding the evidence, engaging customers and ensuring their commitment to the online community is a critical task for the brand to succeed.The online marketing initiatives has put a big challenge on most brands in the aspects of broadcast advertisements, messages, and campaigns by firm marketers that are usually influenced by community partners and ultimately alter the customer's decision beyond the transaction [3]- [5].The postal content generated by marketers via the brand pages are greatly affected by various types of data manipulated through User-Generated Content (UGC) in the form of audio, video, images, animation and text [6].Such activities are the reaction of customers that occur when they are stimulated by the factors around their communities.
The core of the relationship between the customer, the firm, and the brand is through the virtual customer engagement [7]- [8].In particular, there is a demand to provide a broader view of the online community in the social media landscape [9]- [11].The concept of customer engagement is the need for a brand to handle online affairs more effectively and overcome the firm's challenge to understand the exact mechanism to measure online customer preferences and reactions.
Customer engagement is an evolving concept of relationship marketing theory that stands as a key role motivated by various drivers, thereby delivering impact on other consequences (see [12]- [16]).In consumer behaviour research, this principle theory is anchored with servicedominant logic and social exchange theories in which are mutually complementing in terms of the body knowledge contribution.The lens of service-dominant logic has centralized the notion of customer engagement on the value co-creation of firm-customer exchange in the network of interactive service system [17]- [18].According to [19]'s social exchange theory, customers are predicted to respond positively to their thoughts, feelings, and behaviors towards the brand as they are motivated by certain future returns from established obligations [20].Communities interacting with SNS have acknowledged the tendency of customers to be grateful and reciprocate by repaying the benefits they acquired in the process of exchanging with the brand.Dominantly, online customer engagement can be assessed through three main dimensions -cognitive, affective and www.ijacsa.thesai.orgbehaviouralas treated by various researchers in different field areas [10], [21].This study, however, argues about the importance of customer behavior within the SNS community.The behavioral aspect profoundly seen as priority since it serves as a strong indicator of customer engagement in social media and supported by many claims in previous studies [8], [22]- [24].Manifesting behavior from such customers includes all kinds of behaviours that they translate beyond purchase arrangements; such as suggestions for service improvements, product purchasing insights, search of brand and product information, and sharing of usage experience with other customers.Each of their behaviours in the app interface panel greatly determines the involvement of other community members to the brand.
It is clear that the study referred to customer engagement within social media community is increasingly important [2], [25]- [26].These scholars focus on examining the intangible indicators that pose impacts to the engagement domain, for examples reputation, advocacy, and others.In spite of this, literature recognizes the lack of agreement on research of the consequences of customer engagement [21].While the impact of customer engagement in the communities studied is more important as it determines the restructuring of a more effective strategy by the brand for achieving better business objectives, scientific research investigating the behavior of customer engagement in the SNS is rarely implemented [27].In the similar consonant, authors call for more research to gain insights on social media use as an effective integrated marketing tools [28]- [30].The development and verification of indicators to assess the involvement of the community in SNS activities can meet the claims of those researchers.Therefore, to encourage future research on the consequences of customer engagement in behavioural dimension within the SNS community, it is essential to have a systematic approach to validate the indicators.The study seeks to obtain a consensus of experts relating to the development and validity of the proposed consequences of customer engagement due to the scarcity of works conducted to this date.The effective tool that will be used in this case is the Delphi Fuzzy technique in which is expected to work very well as the validity measurement of the variables studied.Thus, the objective of this study is:

What is the level of agreement among experts towards the indicators for evaluating the behavioral manifestation from the engaged customer (consequences) via the SNS platform?
The rest of the article presents the conceptualization of customer engagement in SNS held by this study, followed by the methodology section to elucidate the implementation of Fuzzy Delphi approach to achieve the above objective.

II. CONCEPTUAL FRAMEWORK
Firms must accommodate opportunities to encourage customers in the participation of product review, exchange information, development, and innovations [31]- [32].These activities can be reached when members of the online community continue to communicate in the direction of bilateral in their post-engagement in SNS through the eWOM network.Brand proactive activities will increase the participation of more customers in the community, in turn will add the concept of sharing and dissemination of brand information.Subsequently, customers involved in the SNS platform are believed to have shown certain reactions of behavior as a result of intent to show their involvement.
This study put a debate on the increases of eWOM and the active participation in brand community as the prominent consequences that symbolizes the behavioral of engaged customers in SNS.Furthermore, knowing the consequences are even more important for firms to get valuable insights of what outcomes after customers are considered as engaged person, hence giving input to their online marketing strategies to be reviewed and optimized as to making them parallel with the business goals.The conceptualization of customer engagement and it's consequences in the SNS brand page is illustrated in Fig. 1 below.Positively engaged customers are likely communicating in positive eWOM about firms, brands and products.In adjustment with the context of this study, eWOM was labelled as -SNS environment which create new forms of group conversations in communities, with large numbers of participants participate in different roles and stages, allowing potential and existing customers to positively express statement and exchange information about a brand, product or a firm‖ [33]- [35].
In the literature of marketing and communication, two important dimensions that conceptualize social influences of customers' attitudes and behaviors in the eWOM process are opinion leadership and opinion seeking [35]- [36].The exceptional growth of SNS offers opinion leaders with the ability to strengthen their interpersonal characteristics by motivating other community fellows, hence will encourage the eWOM development [37].Relatively posses lower product knowledge, opinion seekers actively search information and advice in a given category [35].Another important dimension in eWOM is online pass-along behavior [38] in an attention to provide the community SNS members with a useful tool to exchange meaningful information and past worth experiences about brand and product [37].www.ijacsa.thesai.org

B. Participates in Brand Community
The first concept of brand community introduced to the marketing literature is proposed by [39], defined as -a specialized, non-geographically bound community, based on a structured set of social relationships among admirers of a brand‖.The engaged customers are experienced customers with the focal brand, hence, they highly possible to deliver some valuable knowledge-sharing information to advise and to influence community peers essentially by helping each other and fixing each other's problems with the brand [40].The engaged customers' activities will attract and influence enormous potential customers who interest about the brand to participate and constructing larger groups, and performing many important tasks on behalf of the brand [18], [25].
As a quest to universally connecting all benefits of brand communities announced by prior studies, [26] had identified a diverse set of dimensions which is empirically tested to gain insight of brand communities' motivational driversbrand influence, connecting, helping, like-minded discussion, seeking assistance, self-expression and up-to-date information.Adapted from these empirically tested measures, this study aims to test the relationship between customers engaged with brands with the SNS community in which they are located.In this study, the manifestations of customers involved in SNS is confined within a one-dimensional outcome -which is behaviour.Following the exploration of previous works (refer Table I where the constructs of behaviour dimension are marked with *), it is strongly believed that these two consequences could result in significant relationships with the domain of customer engagement using the Delphi Fuzzy method.To achieve results, the manuscript on hand has adjusted 9 items to Increases in eWOM from [37] and [44], as well as 24 items were drawn and coordinated from [26] to Participates in Brand Communities to measure the scale of customer engagement behavior as conceptualized in Fig. 1.

III. METHODOLOGY
Fuzzy Delphi technique has been utilized by this study to validate the conceptual framework of customer engagement in SNS as well as adapted measurement scales.Inspiration of the use of Fuzzy Delphi is derived from the effectiveness of the results in various types of IT/IS research variations presented by [45].This tool is a very helpful method when the level of research requires approval from a group of experts.In addition, the Fuzzy Delphi technique is an interesting method of collective decision-making about the blurred ideas of the alignment of experts' opinions [46].Their responses can be implemented in a short time because of the use of survey methods in parallel with the lower cost and experts can fully express their true response without misinterpretation to ensure perfection and consistency of opinion [47].
The whole methodology is done by dividing the processes into four major stages.The first stage is prior to the Delphi Fuzzy technique is performed where the items to be certified by experts are developed.The second stage refers to the selection of 12 appropriate experts from the background of the academic and business fields.The third stage is related to data collection activities from designated experts within a certain period of time.The final stage is definitely an analysis process in terms of generating significant relationships between the latent variable for review, customer engagement and the two identified consequences.Table II(A) and II(B) illustrates the four levels mentioned.

A. Instrument Development
The formation of a questionnaire instrument was conducted after the identification of two of the leading consequences of SNS's customer engagement (as presented in Table I).In order to meet the criteria of measurement items that are expected to yield significant results, these items are withdrawn from the relevant prior research.The statement of these questions has been modified to conform to the context and environment of the SNS medium to be studied.These 42 questions were reviewed in terms of content and format at the beginning by 3 academicians in the relevant field of social media and business.In conjunction with this, two language specialists, an English language lecturer and another Malay language lecturer were also invited to evaluate the suitability of the item presented.Some improvements have been made to the content of the question with the aim of enhancing the clarity and ease of formatting.Adaptive questions from previous studies with different mediums and samples can affect the reliability and validity of items to be tested in real surveys.Therefore, this present study has taken into account the views of practitioners (in this case the Small Medium Enterprises (SME)) through small-scale structured interview sessions to validate items captured from literature [45].The examined set of items is then properly formatted into the questionnaire as outlined in Table II(A).www.ijacsa.thesai.org

B. Sampling and Selection of Respondents
The expert group required for this study is from those who meet the following criteria: knowledge and experience with issues being examined, ability and willingness to participate, qualification, individual character, comparative ability, communication skills as well as consistent and credible within their respective fields [46], [48].Hence, this study has chosen the purposive sampling through the identification of 12 skilled experts in one or more of the following areas: IT, social media, online governance, marketing, and consumer behaviour.A total of 12 experts were taken from the suggestion of [49] which states that heterogeneous sampling requires between 10 and 35 experts.The experts who come from different profiles and backgrounds are made up of 7 academics and 5 practitioners (refer Table II(A) for the classification).By using email and follow-up via phone calls, they have been invited to take part in the survey and agreement have been dealt with as soon as they give consent.
By scheduling different meetings, a face-to-face approach with these 12 designated experts are carried out to provide an explanation of the items reviewed if any issues arise.The questionnaires are then distributed and the item verification process from each expert is conducted based on the procedures outlined.This level ends with the collection of forms at different times where the items being evaluated are ready to be analyzed using the Fuzzy Delphi method in Microsoft Excel.www.ijacsa.thesai.org

Data analysis
Step 1 The determination of triangular fuzzy numbers

Step 2 The determination of average responses
Next, the average value of each expert's response will be calculated with the following equation ( 1): Step 3

The determination of group consensus
The overall value of the items in each construct is assessed to exceed 75% in order to reach the group agreement [50], by using the formula (3) as the following: Step 5

The identification of defuzzification and ranking of items
To calculate the defuzzification, either of these formulas (i or ii or iii) can be used: i  ii  iii  The resulting score numbers (defuzzification) must be in the range 0 to 1 [56] and the value of alpha-cut must be exceeding 0.5 [52]- [53] to indicate the expert consensus.

C. Data Analysis using Fuzzy Delphi
The analysis was conducted with the aim to prove that the individual item and the overall examined constructs -Customer Engagement and the two consequences namely Increases in eWOM and Participates in Brand Communitiesare the valid model and will be eligible as the measurement items for the actual survey in the following research phase.Overall, the Fuzzy Delphi systematic analysis needs to be through 5 steps to obtain the final decision.
As summarized in Table II(B), data collected in the form of linguistic scales should first be altered to fuzzy triangular numbers [50].This is done for each expert before following the next procedure, determining average responses.Therefore, in the third step, the calculation of the threshold value (-d‖) for each item and construct is calculated to ensure that the decision does not exceed 0.2 to reach an expert agreement [50]- [51].The overall value of items in each construct is then assessed to obtain a minimum of 75% in order to reach the group's consent [50].This analysis ends by calculating the defuzzification value for each item in prioritizing its position according to the construct [52]- [53].Each of the above steps has been presented distinctly in Table II

IV. RESULT
Fuzzy Delphi analysis that has been implemented on 42 items has met the objective with compelling results.Available in Table III are the results obtained for Online Client Engagement in the dimensions of the behaviour as a latent construct under investigation.Boxes marked with colour display values of -d‖ above 0.2, indicating that some experts disagree with the items reviewed.
However, unmarked boxes have a much larger number, concluding that most items are relevant in their view.It has reached a member consensus percentage of more than 75%, and defuzzification score of individual items is greater than 0.5.With the defuzzification scores, the item ranking can be calculated to know the priorities.Counting scores displaying two items that tend to be an expert choice are the similarity of questions that put forward suggestions and positive feedback about the brand initiated by a customer to other prospects or customers within the community they are interacting with (refer Q3 and Q4).www.ijacsa.thesai.orgOn the other hand, an assessment of the 9 items in the eWOM construct returned the unpredictable results initially.This is because nearly half of the threshold item value (-d‖) has exceeded 0.2 indicating the disagreement of experts on some adapted items from the previous context and research setting.To achieve a good overall agreement, this study has released 4 individual items with the lowest percentage of -d‖ value (not reaching 75%).Consequently, the scale supporting the Increase in eWOM measurements is comprised of only 5 items left as these have the percentage of expert consensus for more than 75%, contributing the total consensus of 80%.Refer Table IV.
According to the same table, score numbers (defuzzification) lie between 0 and 1 [56].The value of alphacut is calculated to be greater than 0.5 [52]- [53] to support the consensus of experts on 5 items for the eWOM consequence.Attitudes to conveying brand-related information to their contact network at SNS are the most accepted items by experts who evaluate the question.Online subscribers are also interested in soliciting opinions or suggestions of members of the branded site community about their new favorite product release.www.ijacsa.thesai.orgThe second consequence of Online Customer Engagement, namely Participates in Brand Communities was captured as significance by the findings (refer Table V(A) and Table V(B)).Examined items were calculated with the value of -d‖ ≤ 0.2, proven that all questions are relevant for reaching an agreement 86%.When the selected group of experts evaluates these items, they have confirmed that they are suitable as measuring scales to the studied constructs.
According to the results of defuzzification in both tables, the scores obtained for all items are within the range of 0 to 1 and the alpha-cut values have exceeded 0.5 [52]- [53], [56].The achievement of these values demonstrates the success of obtaining the experts' consent on each item submitted after the analysis.As mentioned earlier, Participates in Brand Communities has adopted the most number of items from literature.Following that, 9 out of 24 questions (leaving 15 items) have been excluded from this construct based on the ranking to minimize and facilitate respondents in answering the questionnaire in the actual survey soon.www.ijacsa.thesai.orgAfter the analysis and modification process, the number of items retained is 29 against 42 original items reviewed in the provided questionnaire by the expert group.Table VI presents the summary of the items investigated after the Fuzzy Delphi analysis by the experts.The findings demonstrate that items obtained and modified from literature has been validated to be tested with the actual sample of brand communities in SNS environment.V. DISCUSSION AND CONCLUSION Taking the example of the analysis has been discussed, Fuzzy Delphi method proved very effective to confirm the modified set of items of literature into a different dimension of the study in terms of platform, object and subject of customer engagement.Indeed, it shows that Increases in eWOM and Participates in Brand Communities is a relevant result of customers engaged in behavioural aspects of the SNS community network.It coincides with the success of previous researcher studies that use Fuzzy Delphi for verification thus determining the priorities of factors or elements on a subject being examined [46]- [47], [56]- [57].
The analyzed items were well-calculated, as evidenced by the average threshold -d‖ ≤ 0.2 [50]- [51] and the average achievement of specialist consensus for proximal customer engagement variables and both consequences, which exceeded 75% [50].The defuzzification scores obtained for all items is in the range of 0 to 1 and alpha-cut values have exceeded 0.5 as mentioned by [52] and [53] that confirms items analyzed are acceptable and contribute to the body of knowledge.The www.ijacsa.thesai.orgfull analysis results conclude that it has met the objective of this study to validate the measurement items for customer engagement and its consequences from the manifestation of their behaviour on the SNS avenue.
The outcome of this study offers two implications via the side of practical and social.First, by knowing the scales of the customer's engagement in the SNS interface such as indicators like, share and comment, digital marketers can establish effective strategies to drive customer or prospect towards promotional activities, campaigns, and brand-run contests.The items agreed by experts to evaluate the increase of customers through eWOM and the active participation in brand communities should be noted by the firms so that their brand can be recognized by the prospect and maintained by existing customers.For example, customers who continuously react to specific product or brand; by liking or sharing tips and uses of a brand are loyal customers that must be preserved by firms because they are valuable assets for brand life sustainability.This situation dictates the concept of relationship marketing that emphasizes those customers who have interacted with and are familiar with the brand become the source of informers to members within the brand's community site.Other theoretical support for customer engagement in brand community activities is derived from social exchange theory, where customers who are satisfied with the quality of their ties from the brand will resonate the sentiment, for example, through sharing of ideas to advance future brand products.
In terms of social implications, SNS software providers can offer a non-charge indicator within a wider radius to firms, capable of capturing the number of active loyal customers who often contribute brand-related information to other customers.This view aligns with the tendency of firms that often use built-in dashboards as they are cheaper and easier to learn on their own [58]- [59].Programmers and social media administrators are also encouraged to provide similar approaches to these firms.With that, virtual marketers who are mostly small-scale businesses can leverage it to measure the effectiveness of their brand marketing efforts in the community.

Fig. 1 .
Fig. 1.Conceptual Framework of Customer Engagement in SNSA.Increases in Electronic Word-of-Mouth (eWOM)Positively engaged customers are likely communicating in positive eWOM about firms, brands and products.In adjustment with the context of this study, eWOM was labelled as -SNS environment which create new forms of group conversations in communities, with large numbers of participants participate in different roles and stages, allowing potential and existing customers to positively express statement and exchange information about a brand, product or a firm‖[33]-[35]. (B). www.ijacsa.thesai.org

TABLE I .
CONSEQUENCES OF CUSTOMER ENGAGEMENT IN SNS, CONCEPTUAL FRAMEWORK OF CUSTOMER ENGAGEMENT IN SNS

TABLE II .
(A).FUZZY DELPHY -STAGES OF INSTRUMENT DEVELOPMENT, SELECTION OF EXPERTS AND DATA COLLECTION, CONCEPTUAL FRAMEWORK OF CUSTOMER ENGAGEMENT IN SNS

TABLE II .
(B).FUZZY DELPHY -STAGES OF DATA ANALYSIS

TABLE III .
VALUE -D‖ OF ITEM, VALUE -D‖ OF CONSTRUCT, % OF EXPERTS CONSENSUS, DEFUZZIFICATION AND ITEM RANKING FOR ONLINE CUSTOMER ENGAGEMENT (Q1 -Q9)

TABLE IV .
VALUE -D‖ OF ITEM, VALUE -D‖ OF CONSTRUCT, % OF EXPERTS CONSENSUS, DEFUZZIFICATION AND ITEM RANKING FOR INCREASE IN ELECTRONIC WORD-OF-MOUTH (EWOM) (Q10 -Q14)

TABLE V .
(A).VALUE -D‖ OF ITEM, VALUE -D‖ OF CONSTRUCT, % OF EXPERTS CONSENSUS, DEFUZZIFICATION AND ITEM RANKING FOR PARTICIPATES IN BRAND COMMUNITIES (Q15 -Q24)

TABLE V .
(B).VALUE -D‖ OF ITEM, VALUE -D‖ OF CONSTRUCT, % OF EXPERTS CONSENSUS, DEFUZZIFICATION AND ITEM RANKING FOR PARTICIPATES IN BRAND COMMUNITIES (Q25 -Q29)

TABLE VI .
THE MODIFIED NUMBER OF ITEMS AFTER FUZZY DELPHI ANALYSIS