Towards Sustainable Internet Service Provision: Analyzing Consumer Preferences through a Hybrid TOPSIS–SEM–Neural Network Framework

Abstract

In our increasingly digital world, grasping consumer preferences for internet service providers (ISPs) is paramount. This study was conducted in Occidental Mindoro, Luzon, the Philippines, and surveyed 280 respondents across ten municipalities with 81 in-person and online questionnaires. The research focused on twelve latent variables: internet speed, assurance, tangibility, responsiveness, reliability, empathy, data privacy, service quality, value-added services, price, customer satisfaction, and customer loyalty. Methods such as Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), Structural Equation Modeling (SEM), and an Artificial Neural Network (ANN) were integrated for the analysis. The study utilized TOPSIS to rank ISPs based on Service Quality (SERVQUAL) parameters, followed by SEM to delve into variable connections affecting preferences, and through as ANN, consumer behavior and loyalty were predicted. The SEM–ANN results revealed that assurance, responsiveness, empathy, and data privacy directly impacted service quality. Moreover, a significant correlation was found between customer satisfaction and service quality, influencing further customer loyalty alongside pricing. By integrating these methodologies, the study provides a comprehensive understanding of ISP preferences and emphasizes service quality as the most significant factor in industry decision-making.

1. Introduction

Emerging advancements in high-speed technologies are reshaping society, allowing many individuals to access the Internet via their smartphones [1]. Reliance on the internet continues to increase, specifically in the academic field as it provides an infrastructure that makes online resources accessible and facilitates distance learning. However, the lack of reliable and high-speed internet access, particularly in rural and underserved communities, limits the ability of the students to participate in online learning which hinders academic growth and leads to educational inequalities [2]. Additionally, the threat of cybercrime and data breaches caused by the inadequate security measures and vulnerabilities of internet service providers (ISPs) is continuously increasing, which significantly affects the academic sector, which not only threatens the privacy and confidentiality of intellectual property, but also puts at risk the integrity of academic institutions [3]. Given the significant flaws of ISPs, comprehending consumer preferences is necessary for enhancing service standards and meeting user demands [2]. In the Philippines, consumer inclinations towards ISPs are influenced by factors like satisfaction levels, willingness to switch providers [3], and the emphasis placed on quality over price. According to [4], there has been a marginal uptick in consumer satisfaction with ISPs. Key considerations include billing processes, communication methods, costs, customer support, and performance [5]. Over the past year, internet usage has skyrocketed, prompting ISPs to enter the market to swiftly meet this surge in demand. In one study [6], ISP revenue surged to USD 10.7 billion in 1998 and was projected to reach USD 37.4 billion by 2003, representing a growth rate of 28% annually. ISPs stand to gain from acquiring accurate insights into their customers’ perceptions of their brand’s service quality; such insights can empower service brand managers to devise effective marketing strategies to secure a competitive edge and long-term profitability.

This study utilized TOPSIS, SEM, and Artificial Neural Network (ANN) models to provide a comprehensive assessment of customers’ preferences for internet service providers (ISPs). TOPSIS, a multi-criteria decision-making method, allows for the evaluation of service quality dimensions, namely assurance, tangibility, responsiveness, reliability, and empathy. Structural Equation Modeling (SEM) allows for the examination of complex relationships between latent variables that significantly influence customer preferences, offering valuable insights into the decision-making process. ANNs, with their predictive capabilities, strengthen the study by forecasting customer choices based on the identified preferences and the integrated service quality evaluation. By combining these methods, the study aimed to fill the gap in existing research by providing a holistic understanding of customer preferences for ISPs. The integration of TOPSIS, SEM, and an ANN allows for a multifaceted analysis, considering various aspects of service quality, decision-making criteria, and predictive modeling. This comprehensive approach ensures that the study offers meaningful insights into consumer choices, which can be beneficial for ISPs in developing targeted strategies and improving their services to better meet customer needs and preferences.

According to [7], the Internet continues to increase in importance and as a necessity in daily life. Currently, digital technologies are now experiencing a global revolution that is transforming and changing society and company processes [8]. Thus, an understanding of ISP competition under path selection is necessary for making well-informed strategic decisions, both immediately and strictly analytically [9]. This study is crucial since it clarifies the details of customer preferences in the Internet Service Provider (ISP) sector, providing insightful information to consumers and regulators alike. The study’s goals included thoroughly investigating the variables influencing consumer decisions, such as pricing schemes, customer service, and service reliability, and evaluating customer satisfaction and switching patterns. By exploring these factors, the study hopes to provide strategic insights to ISPs to boost competitiveness, improve regulatory frameworks, and eventually contribute to a telecommunications landscape that is more focused on consumer needs [9]. Thus, this study plays a role since it may provide insightful information about consumer behavior in the internet service provider (ISP) sector, a vital telecommunications business area. To create strategies, policies, and services that effectively address customers’ changing requirements and wants, legislators, regulatory bodies, ISPs, and other stakeholders must thoroughly understand consumer preferences [10]. Understanding consumer preferences for internet service providers (ISPs) in Occidental Mindoro, the Philippines, holds significant value for stakeholders. This study will allow consumers to assess the strengths and weaknesses of different internet service providers (ISPs), guiding them in making informed choices suitable for their needs and budgets. For internet service providers (ISP), the study can shed light on areas that need improvement, leading to an increase in market share and customer satisfaction. Furthermore, the research methodology, integrating SERVQUAL, TOPSIS, and Artificial Neural Networks, offers a new approach for academics and researchers, promoting advancement, particularly in analyzing consumer preference in developing countries like the Philippines. By investigating consumer preference in Occidental Mindoro’s ISP market, this study can improve the decision-making skills of all involved parties, promoting a more competitive and consumer-centric internet service landscape in this region.

1.1. Theoretical Research Framework

Internet service providers (ISPs) play a vital role in facilitating access to the internet, which is necessary in performing diverse activities. In this study, the researchers included latent variables such as assurance, tangibility, responsiveness, reliability, and empathy, which are the five dimensions of service quality and other factors, namely internet speed, data privacy, value-added services, price, and customer satisfaction, that significantly influence customer loyalty. This study investigated the current performance of ISPs and the factors that positively affect the customer’s preference in the context of choosing internet service providers. Investigating the relationships of these variables provides helpful insights for ISP stakeholders which will pave the way for sustainable internet in the future. (See Figure 1).

1.2. Hypothesis Development and Literature Review

This study hypothesized that these factors are significantly related to the loyalty of customers to their ISPs which are influenced by the behaviors and choices within the ISPs’ ecosystem. Thus, the researchers suggest the following:

Hypothesis 1 (H1): 

There is a significant relationship between internet speed and service quality.

Access to high-speed internet enables individuals to readily explore a vast array of valuable information, facilitating ongoing learning, a deeper comprehension, and international collaboration [11]. This accessibility fosters personal development, stimulates innovation, and drives economic and social advancement on a global scale. Barbero and Antonelli (1995) [12] stressed the importance of understanding user expectations and delivering precise quality of service (QoS) in high-speed data networks. Additionally, [13] underscored the difficulty of objectively evaluating internet quality metrics like speed. Expanding high-speed internet access not only enhances personal development but also fuels innovation, improves educational outcomes, and boosts economic growth [14]. Additionally, it facilitates telemedicine, strengthens social connections, and promotes inclusivity [15,16]. Therefore, it is vital to prioritize improving internet accessibility and quality to harness its full potential for societal advancement [17].

Hypothesis 2 (H2): 

There is a significant relationship between assurance and service quality.

According to [18], assurance involves effectively distributing information across different levels, supported by a proposed framework for efficient information sharing. Expanding on this idea, ref. [19] emphasized the significance of a network services management structure aimed at enhancing the internet model and guaranteeing high quality of service (QoS). In today’s digital environment, it is crucial for customers to feel secure and confident throughout their interactions with internet service providers (ISPs), including safeguarding personal data and overall consumer confidence [16]. According to [18], assurance involves effectively distributing information across different levels, supported by a proposed framework for efficient information sharing. Expanding on this idea, ref. [19] emphasized the significance of a network services management structure aimed at enhancing the internet model and guaranteeing high quality of service (QoS). In today’s digital environment, it is crucial for customers to feel secure and confident throughout their interactions with internet service providers (ISPs), including safeguarding personal data and overall consumer confidence [20]. To achieve this, ISPs must implement robust cybersecurity measures to protect user data from breaches and cyber threats [21]. Furthermore, clear communication about data handling practices and transparent privacy policies can help build trust with consumers [22]. Ensuring reliable and uninterrupted service is equally important, as consistent connectivity directly impacts user satisfaction and confidence. Advanced monitoring and rapid response systems can help mitigate service disruptions and maintain high QoS standards [23]. By prioritizing these aspects, ISPs can enhance user experiences and foster greater trust and reliability in their services [24].

Hypothesis 3 (H3): 

There is a significant relationship between tangibility and service quality.

Hypothesis 4 (H4): 

There is a significant relationship between responsiveness and service quality.

Tangibility is vital in easing consumer apprehension when choosing internet service providers (ISPs) [25]. This significance is highlighted by the tangible aspects incorporated into digital platforms, which have been proven to positively impact the overall perceived quality and satisfaction [25]. Additionally, responsiveness has emerged as a crucial factor, stressing the importance of service availability and promptness in shaping user perceptions and the effectiveness of internet-based services. Insights from [26] further explored the notion of responsiveness in the context of quality of service (QoS) on the internet, emphasizing its pivotal role in enhancing user experiences and facilitating successful interactions with online services. To further improve user satisfaction, ISPs can focus on providing clear and accessible customer support channels, ensuring that users can quickly resolve any issues they encounter [27]. Offering service guarantees and transparent communication about network performance can also reassure consumers about the reliability of their internet service [28,29], and regular updates and proactive maintenance efforts can demonstrate an ISP’s commitment to maintaining high service standards [30]. By addressing both tangibility and responsiveness, ISPs can significantly enhance their service quality and trust from customers.

Hypothesis 5 (H5): 

There is a significant relationship between reliability and service quality.

Ensuring consistent and high-quality service from internet service providers (ISPs) hinges on the reliability of their networks [31]. Kalmanek [32] underscores the significance of fault-tolerant routing and meticulous network planning to bolster this reliability. Wang (2011) [33] stresses the imperative of enhancing service quality and bolstering network resilience to effectively tackle the myriad of challenges confronted by ISPs. In addition, regular updates and maintenance are crucial to keep the network infrastructure robust and secure [34]. Effective monitoring tools and rapid response teams further ensure that any issues are promptly identified and resolved [35]. Collaborating with other ISPs can also enhance network resilience through redundancy and shared resources [36]. ISPs can leverage advanced technologies such as AI and machine learning to predict and mitigate potential network issues before they impact users [37]. Additionally, investing in robust cybersecurity measures is crucial to protecting the network from external threats and ensure continuous service availability [38]. Building strong partnerships with technology vendors and industry experts can also provide ISPs with cutting-edge solutions and best practices. Finally, fostering a culture of continuous improvement and innovation within the organization helps ISPs to stay ahead of emerging challenges and maintain high standards of service delivery [39]. Overall, these strategies collectively ensure a more reliable and high-quality service for end users.

Hypothesis 6 (H6): 

There is a significant relationship between empathy and service quality.

Hypothesis 7 (H7): 

There is a significant relationship between data privacy and service quality.

Hypothesis 8 (H8): 

There is a significant relationship between data privacy and customer satisfaction.

One study [40] that aimed to investigate the connection between empathy and service quality in the context of the airline industry suggested that empathy plays a vital role in improving service quality, so airlines must emphasize training their employees to be more empathetic towards customers. Furthermore, another study [41] revealed that perceived privacy protection positively influences both service quality and customer satisfaction, highlighting that securing customer data leads to a higher perception of service quality and customer satisfaction, highlighting that securing customer data leads to a higher perception of service quality and customer satisfaction. This emphasizes the significance of robust data privacy measures in enhancing the overall customer experience and satisfaction in the airline industry. Airlines should prioritize implementing comprehensive data protection policies and technologies to build trust and confidence among passengers, ultimately contributing to improved service quality and loyalty.

Hypothesis 9 (H9): 

There is a significant relationship between internet speed and customer satisfaction.

Hypothesis 10 (H10): 

There is a significant relationship between service quality and customer satisfaction.

Perceived website performance, which is directly related to internet speed, demonstrated that slow and unresponsive websites result in customer dissatisfaction with the overall experience, emphasizing the significant role of fast internet speed in an online setting [42]. In addition, service quality positively correlates with customer satisfaction; considering that service quality as a key determinant of customer satisfaction, focusing on enhancing service quality to enhance customer satisfaction is crucial [43]. Therefore, prioritizing efforts to enhance service quality is essential in improving customer satisfaction levels [44]. By focusing on both internet speed optimization and service quality enhancement, businesses can effectively elevate the online experience for their customers [23,29], ultimately fostering greater satisfaction and loyalty.

Hypothesis 11 (H11): 

There is a significant relationship between value-added services and customer satisfaction.

According to [45], perceived value, service quality, and customer satisfaction are the characteristics that can be rated based on the cumulative effects of each characteristic on post-purchase intention. This finding suggests that if telecom businesses try to encourage mobile value-added service consumers to have favorable post-purchase intentions, including the intention to recommend or repurchase the service, then improvement and recognized worth must come first [46]. They can assess whether introducing a particular value-added service can convince clients that the offering is “more valuable than it costs,” through the advantages of the product for customers and the fairness of the cost [47]. Users thereby experience the value added to a particular service in addition to using it, elevating the basic use of the service to improve satisfaction [48].

Hypothesis 12 (H12): 

There is a significant relationship between price and customer satisfaction.

According to [49], price is the amount of money or value that customers trade to profit from the ownership of or use of a good or service. According to [50], “the amount of money exchanged for a product and service” is another pricing definition. In addition, costs are a set of values that buyers trade for the advantages of possessing or utilizing a good or service. Moreover, ref. [51] defined pricing as the total value consumers are willing to give up in exchange for the advantages of obtaining or using a good or service. The primary driver of customer happiness and product loyalty is price since the customer carefully considers whether they are receiving the best value for their money compared to the product or service [52]. The physical environment and how people perceive prices impact customer satisfaction [53].

Hypothesis 13 (H13): 

There is a significant relationship between value-added services and customer loyalty.

Other than voice services, mobile value-added services are digital services added to mobile phone networks where the content can be either self-produced by mobile telecom service providers or provided through partnerships with content suppliers for strategic distribution. Games, ringtones, icons, chats, online browsing, SMSs (short message services), vouchers, and electronic transactions are some of these services [54]. They can contribute five values to customers, including those with time-sensitive requirements and arrangements, impulsive wants and choices, entertainment needs, efficiency needs and goals, and mobility demands [55]. The following hypotheses were suggested by this analysis.

Hypothesis 14 (H14): 

There is a significant relationship between price and customer loyalty.

One study [56] emphasized the impact of price promotions on online businesses’ customer satisfaction and found that these promotions affect customers’ purchasing decisions and client satisfaction. The research results validated member cards and mobile application programs as strategies for cultivating customer relationships. A component of the loyalty program for customers to increase client loyalty was running a promotion with discounted prices [57]. These strategies effectively create a sense of value and exclusivity among customers, encouraging repeat purchases and long-term engagement [58]. Moreover, personalized rewards and benefits from loyalty programs enhance customer satisfaction and brand loyalty [59]. Such promotional efforts not only drive immediate sales but also foster enduring customer relationships. By consistently offering financial incentives, businesses can attract new customers and retain existing ones, ultimately leading to increased customer lifetime value and positive word-of-mouth referrals.

Hypothesis 15 (H15): 

There is a significant relationship between customer satisfaction and customer loyalty.

The levels of customer satisfaction have been increasingly determined to be crucial elements in the competitive distinction to increase customer loyalty and retention [60]. It is crucial to advance clients’ loyalty and ensure customer happiness and retention [61]. It is already widely acknowledged that in the hotel sector, quality of service gets people “hooked” on contentment; keeping hold of current clients is just as important as bringing in new ones, by implementing efficient procedures for customer satisfaction and loyalty [62]. High levels of customer satisfaction significantly enhance the business’s competitive advantage by encouraging repeat business and recommendations, while quality service and positive experiences foster strong emotional connections [63]. Effective loyalty programs and exceptional service quality boost retention rates and attract new customers through positive reviews and referrals, making customer satisfaction crucial for long-term success and competitiveness [64].

2. Materials and Methods

2.1. Participants

The researchers executed an online and face-to-face distribution of survey links to different social media platforms, including Messenger, Facebook, and others, using Google Forms and survey forms, and disseminated them from February 2024 to March 2024 to Filipino citizens residing mainly in the province of Occidental Mindoro, Luzon, the Philippines, who had experience in using different internet service providers. Before the survey started, informed consent was obtained from each participant in the study and the study was also approved by the university research ethics committee (FM-RC-24-54). The sample population was chosen using simple random sampling with a minimum sample size of 200 respondents [65]. A total of 280 respondents in Occidental Mindoro answered the online survey questionnaires and in-person surveys that contained 81 questions. It was found that for a structural equation model analysis with ten to fifteen observation variables, a minimum of 200 respondents is recommended [66].

Table 1. Demographic summary of the participants (n = 280).

Characteristics	Category	N	%
Age	Male	137	48.8
	Female	143	51.1
Generation	Generation Z	184	65.7
	Millennial Generation	95	33.9
Educational Background	Elementary Graduate	2	0.7
	High School Graduate	7	2.5
	Senior High School Graduate	96	34.3
	Technical/Vocational Graduate	23	8.2
	Baccalaureate/College Graduate	84	30
	Post-Baccalaureate Graduate	17	6.1
	No Graduate Completed	18	6.4
	Special Education (undergraduate)	25	8.9
	Special Graduate (graduate)	8	2.9
Municipality	Abra de Ilog	24	8.6
	Calintaan	31	11.1
	Lubang	24	8.6
	Magsaysay	26	9.3
	Mamburao	12	4.3
	Paluan	17	6.1
	Rizal	19	6.8
	Sablayan	17	6.1
	San Jose	71	25.4
	Santa Cruz	39	13.9
Monthly Salary	Less than PHP 15,000	159	56.8
	PHP 15,001–PHP 30,000	83	29.6
	PHP 30, 001–PHP 60,000	30	10.7
	PHP 60,001–PHP 75,000	6	2.1
	Above PHP 75,000	2	0.7

shows the descriptive statistics of the 280 participants; 51.1% were female and 48.8% were male. Approximately 66.43% of respondents were aged 11–26 years old (Generation Z) and 34.57% of respondents were aged 27–42 (Millennial generation). A total of 34.3% were in senior high school, followed by baccalaureate or college graduates at 30%; 8.9% were in special education (undergraduate), 8.2% were a technical or vocational graduate, 6.4% had not completed a graduate degree, 6.1% were a post-baccalaureate graduate, 2.9% were a special graduate, 2.5% were a high school graduate, and 0.7% were an elementary graduate. Most respondents were from San Jose (25.4%), followed by Santa Cruz (13.9%), Calintaan (11.1%), Magsaysay (9.3%), Abra de Ilog and Lubang (8.6%), Rizal (6.8%), Paluan and Sablayan (6.1%), and Mamburao (4.3%). About 56.8% of the respondents had a monthly income of less than PHP 15,000, 29.6% of them have a monthly income of PHP 15,001–PHP 30,000, 10.7% of them have a monthly income of PHP 30,001–PHP 60,000, 2.1% of them have an income of PHP 60,001–PHP 75,000 monthly, and 0.7% have a monthly income of more than PHP 75,000.

2.2. Questionnaires

The self-administered questionnaire was created based on the theoretical framework to analyze customer preferences regarding internet service providers. The researchers collected and gathered data using online and in-person surveys, which were all assessed from the user’s point of view. As indicated in the table, the construct variables included in the questionnaire were as follows: (1) internet speed; (2) assurance; (3) tangibility; (4) responsiveness; (5) reliability; (6) empathy; (7) data privacy; (8) service quality; (9) value-added services; (10) price; (11) customer satisfaction; and (12) customer loyalty. The questions were thoughtfully developed to help the researchers understand and analyze consumer preferences for Internet Service Providers. A Likert-type scale with numerical values was utilized to measure the participants’ responses. A Likert-type scale was employed with numerical values (1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree, and 5 = strongly agree). (See

Table 2. The constructs and measurement items.

Construct	Item	Measure	Supporting Ref.
Internet Speed	IS1	My internet speed is slow in downloading large files.	[67]
	IS2	I always experience lagging or buffering when streaming videos or online games.	[68]
	IS3	My internet speed is not consistent in meeting the advertised download and upload speeds.	[69]
	IS4	I am not satisfied with the responsiveness of my internet connection, especially during peak usage times.	[70]
	IS5	I cannot multitask online activities because of internet disruption.	[71]
	IS6	My internet speed is not reliable so I cannot plan online activities effectively.	[72]
	IS7	Overall, I am not satisfied with the speed of my internet connection provided by my current ISP.	[73]
Assurance	A1	My ISP provides efficiency in fixing technical problems and preserving network stability.	[74]
	A2	I have confidence in my ISP’s capacity of protecting my privacy and online data.	[75]
	A3	I have confidence that my ISP will strongly address any vulnerabilities and possible safety risks.	[76]
	A4	My ISP readily offers its clients assistance materials and useful resources to help them learn about online security.	[75]
	A5	I have confidence that I can get in touch with my ISP, ask for help, and get answers quickly.	[77]
	A6	The internet services offered by my ISP are well-known for being trustworthy and safe.	[78]
	A7	In comparison to its rivals, I think my ISP is a reliable and ethical service provider.	[77]
Tangibility	T1	The Internet Service Provider’s (ISP) physical infrastructure (e.g., cables, towers) is well-maintained and in good condition.	[79]
	T2	My Internet Service Provider (ISP) gives clear and informative communication channels (e.g., website, hotline, social media) for updates and technical support.	[73]
	T3	My Internet Service Provider (ISP) provides clear and easily accessible information about its service plans, coverage areas, and pricing.	[80]
	T4	My Internet Service Provider (ISP) gives timely and accurate bills with clear breakdowns of charges and service usage.	[81]
	T5	My Internet Service Provider (ISP) offers a variety of service packages and plans to cater to different needs and budgets.	[82]
	T6	The user interface and functionality of my Internet Service Provider (ISP) online account portal are user-friendly and easy to navigate.	[83]
	T7	I feel valued and respected as a customer by my Internet Service Provider (ISP).	[73]
Responsiveness	R1	My Internet Service Providers address my inquiries and concerns through their hotline or online channels.	[84]
	R2	My Internet Service Provider’s (ISP) technician arrives at my scheduled appointment times for on-site repairs or installations.	[85]
	R3	My Internet Service Provider (ISP) gives about disruption and estimated restoration times through clear and timely communication.	[86]
	R4	My Internet Service Provider (ISP) offers convenient and accessible options for me to report issues and seek technical assistance (e.g., online portal, mobile app).	[87]
	R5	My Internet Service Provider (ISP) resolves Internet service issues efficiently and effectively within a reasonable timeframe.	[88]
	R6	My Internet Service Provider’s (ISP) customer service is readily available and helpful when I call for assistance.	[89]
	R7	Considering all factors, I am confident that my Internet Service Provider (ISP) will quickly address any future concerns I may have.	[90]
Reliability	RL1	My internet connection consistently offers right speeds for uploading and downloading that are stated.	[91]
	RL2	During instances high usage and intense online activity, my internet connection functions good.	[92]
	RL3	I have faith in my ISP to quickly and effectively handle any technical problems and internet failures.	[93]
	RL4	The customer assistance staff are friendly and fast in responding to my inquiries and problems.	[77]
	RL5	I would recommend my current ISP to others based on my experience given its consistency and quality of service.	[94]
	RL6	My ISP gives out straightforward and understandable information, regarding service plans, costs, and any potential restrictions.	[95]
	RL7	I get alerts regularly from my ISP regarding upcoming maintenance periods and possible downtime.	[96]
Empathy	E1	My ISP shows that they really do understand my wants and preferences when it comes to internet usage.	[97]
	E2	My ISP demonstrates a strong sense of customer service by quickly responding to my issues and requests.	[77]
	E3	My ISP provides unique service options and plans to meet the various needs and financial constraints of its clients.	[98]
	E4	The customer assistance personnel at my ISP are helpful, kind, and actively listen to my issues.	[77]
	E5	Each client is given individual consideration by my ISP.	[99]
	E6	The employees of my ISP take the initiative on what is needed to satisfy the demands of customers.	[100]
	E7	I believe my ISP is understanding to my inconvenience and is aware of the psychological toll that internet delays may take.	[98]
Data Privacy	DP1	My Internet Service Provider (ISP) is transparent about the type of data they collect from me.	[101]
	DP2	I have control over the type of data my Internet Service Provider (ISP) collects and store.	[102]
	DP3	Accessing and downloading the data that my Internet Service Provider (ISP) has collected about me is easy.	[103]
	DP4	My Internet Service Provider (ISP) provides clear explanation on how they use my data.	[104]
	DP5	I am confident my ISP takes adequate measures to protect my data from security breaches.	[105]
	DP6	I am informed promptly and transparently in case of a data breach involving my information.	[106]
Service Quality	SQ1	I am satisfied with the speed of my internet connection.	[107]
	SQ2	The network performance of ISP is reliable, with minimal latency and packet loss.	[108]
	SQ3	ISP consistently delivers the promised internet speed without significant fluctuations	[109]
	SQ4	The ISP promptly resolves service issues, minimizing downtime.	[110]
	SQ5	ISP is transparent about service updates, maintenance, and any potential disruptions.	[111]
	SQ6	I am satisfied with the download and upload speeds offered by ISP.	[112]
	SQ7	Overall, I am highly satisfied with the quality of internet service provided by ISP	[113]
Value-Added Services	VAS1	The lack of required VAS options might affect whether I choose to continue with my present ISP or think about moving to a different one.	[114]
	VAS2	My general fulfillment with my ISP’s service has increased because VAS is now provided.	[115]
	VAS3	Compared to competitors who don’t provide comparable services, I am more likely to suggest my ISP to others because of the VAS options.	[115]
	VAS4	The absence of required VAS options might affect whether I choose to continue with my current ISP or consider switching to a new one.	[115]
	VAS5	My ISP offers VAS that are different from those of others and greatly enhance the value of my service contract.	[116]
	VAS6	I would be willing to pay for an internet service package that offers excellent and suitable VAS.	[116]
Price	P1	I believe the internet service provided by the ISP offers good value for the price.	[117]
	P2	The ISP’s prices are competitive compared to other internet service providers in the area.	[117]
	P3	I am satisfied that the pricing structure of the ISP has no hidden fees or unexpected charges.	[118]
	P4	The pricing of the ISP is affordable and aligns well with my budget.	[119]
	P5	The ISP provides appealing discounts or promotions that enhance the value of the service.	[120]
	P6	I find the billing statements from the Internet Service Provider to be accurate and consistent with the agreed-upon pricing.	[118]
	P7	Overall, I am content with the pricing structure and affordability of the Internet Service Provider.	[121]
Customer Satisfaction	CS1	I am satisfied with the response time of Internet Service Provider to customer inquiries or support requests.	[6]
	CS2	Contacting customer support at the Internet Service Provider is easily manageable for me.	[122]
	CS3	The ISP effectively resolves any issues or problems I encounter with my internet service.	[123]
	CS4	The ISP demonstrates an understanding of my specific internet needs as a customer.	[109]
	CS5	The support staff at the ISP are courteous and respectful during interactions.	[124]
	CS6	Information about service updates, changes, and related matters is effectively communicated by the ISP.	[111]
	CS7	Overall, I am highly satisfied with the customer service experience provided by the ISP.	[113]
Customer Loyalty	CL1	I feel loyal to my present internet service provider.	[113]
	CL2	I believe that my internet service provider values and appreciates its consumers.	[125]
	CL3	I’ve had excellent encounters with my internet service provider’s promos and discounts.	[126]
	CL4	I trust my internet service provider’s security and privacy safeguards.	[127]
	CL5	I’m more inclined to stick with my existing internet service provider because of the loyalty prizes or programs they provide.	[113]
	CL6	I am willing to pay a premium for the level of service offered by my existing internet service provider.	[128]

).

2.3. Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS)

The Technique for Order Preference by Similarity to an Ideal Solution is one of the most common multiple criteria decision analyses used in many research studies, which was proposed in [129]. Its function is to determine the finest option that is farthest from the negative ideal solution and closest to the positive ideal solution based on its purpose [130]. In the current study, the TOPSIS was used to assess and rank the six internet service providers (ISPs) in Occidental Mindoro, the Philippines: Converge, PLDT, One Sky, Smart Bro Home Wifi, DITO Telecommunity, and Globe Broadband. The TOPSIS was integrated with the SERVQUAL dimensions, such as assurance, empathy, reliability, responsiveness, and tangibility, as pre-defined criteria. Additionally, the TOPSIS aimed to determine and identify the relative performance of each provider. This method is crucial for discerning which SERVQUAL dimensions ISPs may be lacking in, thus aiding in identifying areas for improvement within the industry.

Detailed Process of TOPSIS

The first step in the TOPSIS process is to create an evaluation matrix with m alternatives, n criteria (with weights), and the satisfaction values that correspond to each. The functions’ criteria can be cost functions (less is better) or benefit functions (more is better). The matrix resulting from the intersection of each alternative and criteria, denoted as $X_{ij}$, is represented as ${{(X}_{ij})}_{m x n}$. Subsequently, this matrix undergoes normalization through a specific normalization technique: $r_{ij}=\frac{X_{ij}}{\sqrt{{\sum }_{i=1}^m{X^2}_{ij}}}$, i = 1, 2 …, m and j = 1, 2 …, n. The normalization process converts the different attribute dimensions into non-dimensional attributes, enabling comparisons across various criteria. The normalized values are then multiplied by the corresponding criteria weights ${(r}_{ij}\times w_j)$ to create the weighted normalized matrix $v_{ij}$. Following this, the ideal positive solution $A^{+}$ and the ideal negative solution $A^{-}$ are identified based on the weighted normalized values.

$$\begin{gathered} A^{+}=\left\{\left(max{ v}_{ij}|j \in J\right),\left(min{v}_{ij}|j \in J\right)|i=\mathrm{1,2},\dots m\right\} \\ =\{v_1^{+},v_2^{+},\dots , v_j^{+},\dots , v_n^{+}\} \\ {A}^{-}=\{\left(min{v}_{ij}|j \in J\right), (max{v}_{ij}|j \in J)|i=\mathrm{1,2},\dots m\} \\ =\{v_1^{-},v_2^{-},\dots , v_j^{-},\dots , v_n^{-}\} \end{gathered}$$

where

$$J = \{j = 1, 2,\dots , n|j associated with benefit criteria\}$$

$$J = \{j = 1, 2,\dots ,n|j associated with cost criteria\}$$

While the negative ideal solution maximizes the cost criteria and minimizes the benefit criteria, the ideal positive solution maximizes the benefit criteria and minimizes the cost criteria. Next, each alternative’s separation from these solutions is computed. The following are the positive ideal separation ${S_i}^{+}$ and negative ideal separation ${S_i}^{-}$:

$$\begin{gathered} S_i^{+}=\sqrt{\sum _{j=1}^n{(v_{ij}-v_j^{+})}^2} i = 1, 2, \dots , m \\ {S}_i^{-}=\sqrt{\sum _{j=1}^n{(v_{ij}-v_j^{-})}^2} i = 1, 2,\dots , m \end{gathered}$$

The alternatives are ranked based on their relative closeness $C_i$ to the ideal solution $A^{+}$. The alternatives can be arranged in descending order according to the value of $C_i$, with the alternative closest to 1 being the preferred choice. This ranking allows the decision maker to identify the alternative that is closest to the ideal solution, considering both the distance from the positive ideal solution and the negative ideal solution.

$$\begin{gathered} C_i=\frac{S_i}{S_i^{+}+S_i^{-}}, 0<C_i<1, i = 1, 2, \dots ,m \\ {C}_i=1 if A_i=A^{+} \\ {C}_i=0 if A_i=A^{-} \end{gathered}$$

2.4. Structural Equation Modeling

SEM is a technique that combines factor analysis and regression to examine and analyze multivariate relationships between latent variables; it can be applied to analyze a collection of hypotheses from an entire model [131]. Additionally, it is a multivariate technique that is commonly applied, mainly in social and behavioral science, for testing and simulating model hypotheses [132]. This study used SEM to test twelve latent variables: internet speed, assurance, tangibility, responsiveness, reliability, empathy, data privacy, service quality, value-added services, price, customer satisfaction, and customer loyalty.

2.5. Artificial Neural Network

Artificial Neural Networks (ANNs) have become a powerful tool for a wide range of applications [133]. Artificial Neural Networks (ANNs) improve their accuracy by continuously learning from training data over time. ANNs have been used in many research studies to forecast and analyze global human behavior [134]. The Structural Equation Modeling (SEM) approach applied in this study could over simplify and yield unexpected results [135]. Thus, this model was used to determine the most effective activation function for the hidden and output layers, the number of nodes, the optimizer, and the ratios of training and testing to achieved the optimal results.

3. Results

3.1. TOPSIS Results

The descriptive statistics of the participants, as shown in

Table 3. Descriptive statistics of participants (n = 280).

Characteristic	Category	N	%
Generation	Generation Z	186	66.44
	Millennial	94	33.57

, showed that among the 280 participants, 66.43% belonged to Generation Z, while 34.57% belonged to the Millennial generation.

Table 4. Generation Z (age 11–26).

ISP	Dev from S+	Ranking
Converge	0.3841	3
PLDT	0.0806	5
One Sky	0.1124	4
Smart Bro Home Wifi	0	6
DITO Telecommunity	1	1
Globe Broadband	0.5615	2

indicates that DITO Telecommunity was Generation Z’s favored internet service provider, with Smart Bro Home WIFI ranking the lowest in preference.

Based on

Table 5. Millennials (aged 27–42).

ISP	Dev from S+	Ranking
Converge	0.1064	4
PLDT	0	5
One Sky	0.1811	3
Smart Bro Home Wifi	0	5
DITO Telecommunity	0.9203	1
Globe Broadband	0.9053	2

, DITO Telecommunity emerged as the preferred internet service provider among Millennials, while PLDT and Smart Bro Home Wifi ranked lowest in preference.

Table 6. Generation Z and Millennials.

ISP	Dev from S+	Ranking
Converge	0.2567	3
PLDT	0.0323	5
One Sky	0.1231	4
Smart Bro Home Wifi	0	6
DITO Telecommunity	1	1
Globe Broadband	0.7326	2

demonstrate that DITO Telecommunity was the most preferred internet service provider among the 280 consumers, whereas Smart Bro Home Wifi was the least preferred.

3.2. SEM Results

Structural Equation Modeling (SEM) provides advantages over the traditional techniques used in analyzing complex models [136]. Figure 2 demonstrates the initial SEM for the factors influencing the service quality of and loyalty to internet service providers.

This study examined the fundamental factors influencing customer preference for internet service providers (ISPs). Twelve latent variables were investigated using a structural equation model (SEM) to establish a connection between assurance (A), tangibility (T), responsiveness (RS), reliability (RL), empathy (E), internet speed (IS), data privacy (DP), service quality (SQ), value-added services (VAS), price (P), customer satisfaction (CS), and customer loyalty (CL). The SEM results showed that out of the 15 hypotheses, 7 displayed insignificant relationships, namely internet speed, tangibility, and reliability towards service quality; data privacy, internet speed, and value-added services towards customer satisfaction; and value-added services towards customer loyalty, having a p-value above 0.05 which does not meet the standard cut off criteria of SEM. Hence, a revised SEM was obtained by removing the insignificant hypotheses, following a previous study [137], as displayed in Figure 3.

Table 7. Descriptive statistics results.

Factor	Item	Mean	St Dev	Factor Loading
				Initial Model	Final Model
Internet Speed	IS1	3.964	0.7659	0.801	-
	IS2	4.229	0.8988	0.775	-
	IS3	4.129	0.8239	0.712	-
	IS4	4.161	0.8712	0.773	-
	IS5	4.225	0.8693	0.712	-
	IS6	4.111	0.8584	0.805	-
	IS7	4.075	0.9147	0.822	-
Assurance	A1	3.968	0.8137	0.772	0.772
	A2	4.064	0.9096	0.740	0.742
	A3	3.961	0.9051	0.833	0.835
	A4	4.021	0.9037	0.756	0.743
	A5	4.111	0.9189	0.834	0.836
	A6	4.150	0.8994	0.825	0.816
	A7	4.111	0.9110	0.822	0.827
Tangibility	T1	4.039	0.8045	0.820	-
	T2	4.039	0.8687	0.724	-
	T3	4.025	0.8937	0.790	-
	T4	4.043	0.8910	0.736	-
	T5	4.111	0.8542	0.787	-
	T6	4.146	0.8572	0.765	-
	T7	4.171	0.8671	0.795	-
Responsiveness	R1	4.029	0.8546	0.804	0.805
	R2	3.989	0.8980	0.780	0.781
	R3	3.964	0.8585	0.786	0.786
	R4	4.064	0.8569	0.738	0.738
	R5	4.046	0.8217	0.732	0.730
	R6	4.046	0.8601	0.775	0.773
	R7	4.111	0.8625	0.838	0.839
Reliability	RL1	3.986	0.8550	0.803	-
	RL2	3.982	0.8817	0.754	-
	RL3	3.943	0.8490	0.741	-
	RL4	4.025	0.8442	0.700	-
	RL5	4.075	0.8581	0.787	-
	RL6	4.121	0.8714	0.854	-
	RL7	4.064	0.8442	0.790	-
Empathy	E1	4.029	0.8160	0.782	0.791
	E2	4.089	0.8567	0.809	0.837
	E3	4.054	0.8169	0.716	0.714
	E4	4.150	0.8628	0.781	0.784
	E5	4.179	0.8489	0.777	0.755
	E6	4.039	0.8479	0.771	0.731
	E7	4.043	0.8788	0.818	0.822
Data Privacy	DP1	3.996	0.8360	0.789	0.782
	DP2	4.096	0.8643	0.758	0.757
	DP3	3.882	0.8574	0.794	0.793
	DP4	4.021	0.8837	0.737	0.756
	DP5	3.986	0.8675	0.767	0.764
	DP6	4.046	0.9282	0.822	0.835
Service Quality	SQ1	3.982	0.8777	0.654	0.646
	SQ2	4.043	0.9187	0.628	0.612
	SQ3	4.000	0.8920	0.607	0.596
	SQ4	4.021	0.8673	0.667	0.654
	SQ5	4.114	0.8558	0.620	0.606
	SQ6	4.104	0.8634	0.615	0.603
	SQ7	4.246	0.8296	0.522	0.509
Value-Added Services	VAS1	3.918	0.8057	0.758	-
	VAS2	4.025	0.8977	0.770	-
	VAS3	3.957	0.8413	0.748	-
	VAS4	3.964	0.8832	0.741	-
	VAS5	4.046	0.8261	0.761	-
	VAS6	4.118	0.8189	0.790	-
Price	P1	3.943	0.8861	0.798	0.800
	P2	4.032	0.8566	0.766	0.764
	P3	3.996	0.8778	0.766	0.764
	P4	4.089	0.8692	0.766	0.763
	P5	4.032	0.9054	0.839	0.848
	P6	4.011	0.8737	0.749	0.762
	P7	4.043	0.9069	0.780	0.776
Customer Satisfaction	CS1	4.054	0.8169	0.562	0.598
	CS2	4.039	0.8646	0.526	0.560
	CS3	4.014	0.8757	0.520	0.559
	CS4	4.068	0.8793	0.566	0.601
	CS5	4.068	0.8587	0.562	0.593
	CS6	4.018	0.8052	0.564	0.595
	CS7	4.121	0.8673	0.616	0.651
Customer Loyalty	CL1	4.011	0.8402	0.556	0.550
	CL2	4.068	0.8070	0.512	0.513
	CL3	3.968	0.8482	0.558	0.554
	CL4	4.050	0.8661	0.545	0.544
	CL5	4.021	0.8379	0.466	-
	CL6	4.129	0.9301	0.533	0.529

shows the loading value of both the initial and final structural equation models. The factor loadings represent the connection between each indicator and the latent variable under investigation. A higher factor loading, approaching 1.0, signifies a stronger correlation between the indicator and the underlying construct. The factor loading of the initial SEM ranged from 0.466 to 0.854. Following the methodology of a previous study [132], indicators with factor loading values below 0.5 were excluded from the final model. Specifically, CL5 had a loading value of 0.466. Additionally, latent variables, namely tangibility, reliability, internet speed, and value-added services, were eliminated from the final model as we failed to establish a significant relationship with the remaining variables. This refinement improved the overall fit of the model. The remaining indicators in the final model exhibited moderate to strong relationships with the proposed hypotheses, with factor loadings ranging from 0.509 to 0.839. While some factor loadings increased or remained stable between the initial and final models, others decreased.

Table 8. Summary of hypotheses.

	Hypothesis	p	Interpretation
H1	There is a significant relationship between internet speed and service quality.	0.105	Not Significant
H2	There is a significant relationship between assurance and service quality.	0.003	Significant
H3	There is a significant relationship between tangibility and service quality.	0.603	Not Significant
H4	There is a significant relationship between responsiveness and service quality.	0.004	Significant
H5	There is a significant relationship between reliability and service quality.	0.152	Not Significant
H6	There is a significant relationship between empathy and service quality	0.009	Significant
H7	There is a significant relationship between data privacy and service quality.	0.001	Significant
H8	There is a significant relationship between data privacy and customer satisfaction.	0.933	Not Significant
H9	There is a significant relationship between internet speed and customer satisfaction.	0.085	Not Significant
H10	There is a significant relationship between service quality and customer satisfaction.	0.004	Significant
H11	There is a significant relationship between value-added services and customer satisfaction.	0.113	Not Significant
H12	There is a significant relationship between price and customer satisfaction.	0.001	Significant
H13	There is a significant relationship between value-added services and customer loyalty.	0.191	Not Significant
H14	There is a significant relationship between price and customer loyalty.	0.031	Significant
H15	There is a significant relationship between customer satisfaction and customer loyalty.	0.001	Significant

illustrates the significant relationships, indicated by p-values below 0.05, between various constructs: assurance and service quality (H2), responsiveness and service quality (H4), empathy and service quality (H6), data privacy and service quality (H7), service quality and customer satisfaction (H10), price and customer satisfaction (H12), price and customer loyalty (H14), and customer satisfaction and customer loyalty (H15). Conversely, non-significant relationships, with p-values above 0.05, were observed between internet speed and service quality (H1), tangibility and service quality (H3), reliability and service quality (H5), data privacy and customer satisfaction (H8), internet speed and customer satisfaction (H9), value-added services and customer satisfaction (H11), and value-added services and customer loyalty (H13).

Table 9. Accepted fit values.

Goodness-of-Fit Measures for SEM	Minimum Cutoff	Reference
CMIN/DF	<3.0	[138]
Comparative Fit Index (CFI)	>0.7	[139]
Incremental Fit Index (IFI)	>0.7	[136]
Tucker–Lewis Index (TLI)	>0.7	[140]
Root Mean Square Error (RMSEA)	<0.1	[141]

indicates the accepted fit values. These indices are significant in evaluating the goodness of fit of a model in SEM, which also implies that the level of complexity is reasonable; thus, it ensures the accuracy, generalizability, validity, and reliability of the results of the statistical model used.

Table 10. Model of validity.

Factor	Number of Items	Cronbach’s α	Composite Reliability	Average Variance Extracted
Assurance	7	0.994	0.920	0.622
Responsiveness	7	0.915	0.916	0.609
Empathy	7	0.915	0.908	0.585
Data Privacy	6	0.902	0.896	0.589
Service Quality	7	0.920	0.914	0.605
Privacy	7	0.916	0.911	0.594
Customer Satisfaction	7	0.917	0.915	0.607
Customer Loyalty	5	0.885	0.885	0.607

presents the outcomes concerning the reliability and convergent validity constructs through the Structural Equation Modeling (SEM) approach. The examination of reliability and convergent validity utilized Cronbach’s alpha, Composite Reliability (CR), and Average Variance Extracted (AVE). The AVE was calculated by considering the mean value of the outer variables in relation to their association with the corresponding constructs. Additionally, in accordance with the recommendation in [141], advocating for a convergent validity threshold exceeding 0.5 for the AVE, the researchers ensured compliance with this criterion. Consequently, it was noted that the accumulated values surpassed this threshold, indicating the consistency, validity, and reliability of the items within the sample.

In this context, CMIN/DF registered at 2.983, with CFI at 0.082, IFI at 0.803, TLI at 0.792, and RMSEA at 0.084. RMSEA fell slightly short of the recommended threshold, indicating a marginal fit [142]. However, there is room for improvement since [114] suggested that an CFI exceeding the suggested threshold of 0.70 highlights a potential avenue for enhancement, as shown in

Table 11. Model fit.

Goodness of Fit Measures for SEM	Parameter Estimate	Minimum Cut-Off	Interpretation
CMIN/DF	2.983	<3.0	Acceptable
Comparative Fit Index (CFI)	0.802	>0.7	Acceptable
Incremental Fit Index (IFI)	0.803	>0.7	Acceptable
Tucker–Lewis Index (TLI)	0.792	>0.7	Acceptable
Root Mean Square Error (RMSEA)	0.084	>0.1	Acceptable

.

Table 12. Direct, indirect and total effects.

No.	Variables	Direct Effect	p-Value	Indirect Effect	p-Value	Total Effect	p-Value
1	SQ → DP	0.731	0.001	-	-	0.731	0.001
2	SQ → E	0.245	0.007	-	-	0.245	0.007
3	SQ → RS	0.447	0.004	-	-	0.447	0.004
4	SQ → A	0.402	0.006	-	-	0.402	0.006
5	SQ → P	-	-	-	-	-	-
6	SQ → CS	-	-	-	-	-	-
7	SQ → CL	-	-	-	-	-	-
8	CS → DP	-	-	0.505	0.001	0.505	0.001
9	CS → E	-	-	0.169	0.01	0.169	0.01
10	CS → RS	-	-	0.309	0.003	0.309	0.003
11	CS → A	-	-	0.278	0.005	0.278	0.005
12	CS → P	0.731	0.001	-	-	0.731	0.001
13	CS → SQ	0.691	0.004	-	-	0.691	0.004
14	CS → CL	-	-	-	-	-	-
15	CL → DP	-	-	0.512	0.001	0.512	0.001
16	CL → E	-	-	0.171	0.01	0.171	0.01
17	CL → RS	-	-	0.313	0.003	0.313	0.003
18	CL → A	-	-	0.281	0.005	0.281	0.005
19	CL → P	-	-	0.741	-	0.741	0.001
20	CL → SQ	-	-	0.7	-	0.7	0.004
21	CL → CS	1.014	0.002	-	-	1.014	0.002

shows the causal relationships between the variables. The table illustrates that every connection with a probability value less than 0.05 is significant. All of the table’s columns provide important statistical data, most notably the associated p-values and data on the direct, indirect, and total effects of one variable on another. Thus, the statistical significance of the correlations between the variables in the model is illustrated.

3.3. Artificial Neural Network Results

Figure 4 illustrates the architecture of the Artificial Neural Network (ANN). The Structural Equation Modeling (SEM) analysis indicated a positive correlation between all latent variables and each connection, with p-values exceeding 0.05. Likewise, in the ANN framework, co-variances such as A, RS, E, DP, SQ, P, CS, and CL served as input nodes, demonstrating significant correlations with the study’s endpoint variable, customer loyalty. The ANN results highlighted SQ as the most influential factor, followed by DP, A, CS, and P. While both the SEM and ANN yielded consistent findings, it is crucial to recognize the indirect effects of SEM, as they can substantially influence the SEM outcomes, as noted in prior research [143].

Table 13. RMSE values.

Neural Network	Training Dataset		Testing Dataset
	80% Data Sample 275, n = 220		20% Data Sample 275, n = 55
	SSE	RMSE	SSE	RMSE
ANN1	0.526	0.0484	0.107	0.0437
ANN2	0.703	0.0560	0.078	0.0559
ANN3	0.395	0.0419	0.814	0.0373
ANN4	0.324	0.0380	0.064	0.0338
ANN5	0.578	0.0386	0.098	0.0584
ANN6	0.494	0.0469	0.063	0.0418
ANN7	0.482	0.0463	0.271	0.0695
ANN8	0.671	0.0547	0.054	0.0310
ANN9	0.516	0.0479	0.186	0.0555
ANN10	0.412	0.0428	0.101	0.0576
	Mean	0.0461	Mean	0.0484

displays the training and testing datasets across ten trials. The researchers utilized these trials to identify the minimum root mean square error (RMSE) in the Artificial Neural Network (ANN), ensuring that it fell below the root mean square error of approximation (RMSEA) for Structural Equation Modeling (SEM). The findings revealed that the mean RMSE of the training dataset, at 0.0461, and the testing dataset, at 0.0484, both fell below the threshold of 0.084. Consequently, the ANN results were deemed acceptable and accurate. Moreover, Figure 5 translates the values in Table 13 into a line graph to illustrate the result.

Figure 6 illustrates a relative importance graph, indicating the latent variables with the strongest correlation to the dependent variable, or the endpoint variable, which in this study represents customer loyalty.

Table 14. Relative importance.

Predictor (Independent Variable)	Average Relative Importance	Normalized Importance (%)	Rating
Assurance	0.172	40.8%	3
DP	0.228	54.1%	2
SQ	0.421	100.0%	1
P	0.057	13.4%	5
CS	0.122	29.0%	4

presents the independent variable importance score to verify that the results of both models were the same, which implies that the findings are significant and were verified by the SEM–ANN hybrid.

4. Discussion

The study focused on analyzing consumer preferences in terms of internet service providers. An MCDA was employed in this study, particularly the TOPSIS, which was used to rank six (6) internet service providers (ISP), namely Converge (ISP1), PLDT (ISP2), One Sky (ISP3), Smart Bro Home Wifi (ISP4), DITO Telecommunity (ISP5), and Globe Broadband (ISP6), with the use of five dimensions, tangibility, responsiveness, reliability, assurance, and empathy, within the ten municipalities of the province of Occidental Mindoro: San Jose, Santa Cruz, Calintaan, Magsaysay, Abra de Ilog, Lubang, Rizal, Paluan, Sablayan, and Mamburao. A total of 280 respondents were obtained through online and face-to-face surveys. The population was categorized into two generations, and the primary respondents of the study were from Generation Z and the Millennial generation. The results were obtained through the ranking of the internet service providers and revealed that Smart Bro Home Wifi (ISP4) was in the lowest rank; for ISP4, the provider needs to focus on improving and better facilitating the dimensions mentioned.

SEM was utilized to investigate the relationship between assurance (A), tangibility (T), responsiveness (RS), reliability (RL), empathy (E), internet speed (IS), data privacy (DP), service quality (SQ), value-added services (VAS), price (P), customer satisfaction (CS), and customer loyalty (CL). The SEM results showed that service quality was significantly influenced by assurance (β: 0.402, p = 0.006), responsiveness (β: 0.447, p = 0.004), and empathy (β: 0.245, p = 0.007). Similarly, ref. [144] applied the SERVQUAL model to investigate the five dimensions of service quality: assurance, empathy, tangibility, reliability, and responsiveness. The findings revealed that the SERVQUAL dimensions significantly affected how customers perceive service quality. However, the current study failed to establish a relationship between tangibility (p = 0.603) and reliability (p = 0.152) in measuring the services provided by ISPs in the Philippines, which suggests further exploration is needed. Furthermore, data privacy had a significant relationship with service quality, which implies that service providers must emphasize the proper regulation of the data provided by customers, as it affects how they perceive the quality of the services offered. This finding is supported by [69], which concluded that when customers received and felt that the provided services are appropriately given, it resulted in a higher perception of service quality and customer satisfaction. In addition, service quality was positively correlated with customer satisfaction (β: 0.691, p = 0.004), which implies that the ability of the internet service provider to provide service met or surpassed the demands and expectations of the consumer, which is crucial as it positively influences customer satisfaction. Aligning with this finding, ref. [145] found that service quality dimensions in the context of hotel services positively influence customer satisfaction, suggesting that service providers should emphasize the service quality dimensions as they significantly affect their business image and operations. Moreover, the price, which is defined as the amount that was calculated based on numerous factors, had a direct significant effect on customer satisfaction (β: 0.731, p = 0.001), which suggests that consumers were more likely to be satisfied if the pricing of the product or services was close to their preferred amount. One study [146] affirmed that customers are more likely to feel satisfied if the value of the purchased goods and services aligns with their cost. There was also a significant correlation between customer satisfaction and loyalty (β: 1.014, p = 0.002), indicating that customers who feel satisfied with the provided services display a positive intention to purchase frequently and recommend the services to others. Successful businesses must prioritize satisfying customers as it drives customer loyalty [141]. However, the SEM results showed that internet speed did not significantly influence service quality (p = 0.105) and customer satisfaction (p = 0.085), which implies that the ability of an ISP to provide a fast internet connection does significantly affect the customer’s perception of the quality of the service offered and their satisfaction. In contrast, ref. [147] stated that internet speed is a fundamental factor influencing customers’ perception of the service, affecting customer satisfaction and their intention to switch to a different service provider. Furthermore, customer satisfaction was not directly influenced by data privacy (p = 0.933) and value-added services (p = 0.113), which indicates that the ability of service providers to handle the data correctly and elevate the customer experience by providing additional offerings does not contribute to customer satisfaction. In contrast, previous studies claimed that securing customer data leads to higher customer satisfaction [53], and service providers offering value-added services were more likely to satisfy their customers and improve their experiences [148]. Lastly, there seemed to be a more significant connection between value-added services and customer loyalty (p = 0.191), which suggests that additional services provided by the industry will not guarantee that customers will stay committed to the service provider. Zhang et al. (2023) [47] opposed this finding and indicated that providing additional value beyond fundamental products or services is a strategy to foster long-lasting customer loyalty.

Artificial Neural Networks (ANNs) were utilized to establish accurate results and reliable findings to further explore the factors that significantly influence consumer preferences for internet service providers (ISPs). Within this framework, service quality was the paramount factor influencing consumer choices. This was evidenced by its substantial average relative importance score of 0.882, signifying its pivotal role in shaping consumer behavior. Service quality was closely followed by data privacy, which garnered an average relative importance of 0.572. Thus, this highlighted the increasing importance of safeguarding personal information is to consumers in today’s digital landscape. Assurance and customer satisfaction followed with relative importance scores of 0.172 and 0.122. This study found that while factors like assurance and customer satisfaction were important, they played a secondary role compared to the primary factors in shaping consumer decision-making processes. Additionally, privacy was the least influential factor, with an average relative importance of 0.057. Although it may seem less important than other factors, its existence highlights that consumers are consistently concerned about safeguarding their privacy rights. The combination of SEM and ANN methodologies offers a comprehensive analytical approach. The SEM enabled the exploration of causal relationships among the variables, while the ANN’s nonlinear nature enhanced the accuracy and reliability of the results. By integrating the strengths of both methodologies, the hybrid SEM–ANN approach yielded the optimal outcomes, as it was more effective than using any of these methods individually. This approach is a valuable and effective way to analyze and understand consumer preferences in the changing world of internet services.

Theoretical Contributions

According to the “2020 Census of Population and Housing: Age and Sex Distribution in the Philippine Population”, the Philippines has a total population of 109,035,343. Age groups were used to split the population into groups that corresponded to distinct generations. According to the census, there were 9,362,325 people in Generation Z, or those between the ages of 5 and 24. There were 11,374,597 people in the Millennial age group, or those aged 15–30. Thus, there were 12,433,053 people between the ages of 15 and 64, and Occidental Mindoro had a total population of 525,354 individuals [149]. Using simple random sampling, the online survey and in-person questionnaires contained 81 questions were answered by 280 respondents, specifically people from Generation Z and the Millennial generation. Using SEM, the associations between the eleven latent variables were examined. The data from multiple relevant research projects were used to develop the questionnaire through online and in-person surveys. The researchers checked and removed duplicate responses, ensuring that the sample matched the demographic requirements of the study. This method enhanced the overall data collection process, resulting in higher-quality data and potentially reducing the need for extensive data filtering.

The TOPSIS results pinpointed Smart Bro Home Wifi (ISP4) as requiring significant improvements across various dimensions to meet consumer expectations, thereby providing a clear roadmap for enhancing their competitiveness in the market. The SEM findings further underscored the crucial role of dimensions such as assurance, responsiveness, and empathy in positively influencing service quality and, consequently, customer satisfaction. Notably, the SEM highlighted the crucial need to address data privacy concerns to enhance service quality perception, emphasizing its impact on consumer satisfaction and loyalty. Moreover, the direct correlation between service quality, customer satisfaction, and loyalty reinforced the importance of meeting consumer expectations to foster positive word-of-mouth and sustained loyalty. The integration of an Artificial Neural Network (ANN) alongside SEM revealed service quality as the most influential factor shaping consumer behavior, followed by data privacy, assurance, and customer satisfaction. The ANN analysis not only strengthened the significance of these factors but also shed light on consumers’ increasing concerns about safeguarding personal information in the digital age, thereby enriching the understanding of evolving consumer preferences and decision-making processes.

5. Conclusions

Internet service providers (ISPs) are essential in facilitating access to information, communication, economic development, education, and entertainment, underscoring the importance of understanding consumer preferences for ISPs. As the key players directly delivering the internet to consumers, understanding consumer preferences for ISPs is crucial, as they can significantly impact the overall consumer experience. From the streaming of several of our shows to social interactions via social network sites, every facet of our existence involves the services of companies such as ISPs. The Internet’s capabilities depend on the connection we obtain from our internet service provider (ISP) and the speed at which a connection is formed, enabling one to access online information quickly. Further, ISPs have absolute reign in controlling what can be seen and how much data are transferred, depending on the agreed terms and conditions. This aspect underscores the importance of regulations such as net neutrality, emphasizing the essential need for their implementation. In addition, addressing the security and privacy of personal data is vital, requiring ISPs to be transparent and proactive in handling the user information flowing through their networks. Despite challenges and issues, ISPs are at the forefront of innovating new technologies to improve the broadband infrastructure for faster internet connections. In essence, ISPs are essential in shaping our daily online experiences, significantly influencing how we access and interact with internet content. By tailoring services to meet local needs and expectations, ISPs can enhance customer satisfaction and loyalty while making decisions about infrastructure investments and service expansions to contribute to regional development. In this study, the researchers employed digital and traditional methods to gather data from 280 participants, using a comprehensive questionnaire consisting of 81 questions covering 12 factors. The Structural Equation Modeling (SEM) analysis revealed significant correlations between several factors affecting service quality, such as assurance, responsiveness, empathy, and data privacy, which are essential in shaping customer satisfaction. Moreover, the analysis highlighted the significant associations between customer satisfaction and critical determinants like service quality and price. Furthermore, it emphasized the noteworthy impact of price on customer loyalty. Assurance was the most essential factor in evaluating service quality, with service quality and customer satisfaction also playing significant roles. Responsiveness and empathy also played roles as they exhibited relatively minor impacts.

Furthermore, applying the TOPSIS method showed the Philippines’ ranking of internet service providers (ISPs). DITO Telecommunity came in first, and the results showed that it offers fast and reliable internet connection within the municipalities. Globe Broadband was second, Converge was third, One Sky was fourth, PLDT was fifth, and Smart Bro Home Wifi was sixth. Thus, this showed that DITO Telecommunity has stayed relevant in a fast-paced market and can ensure customer satisfaction through its services. This, along with other reasons, makes it an excellent service with affordable prices among ISPs, thus making it the number one choice among telecommunication services within the province of Occidental Mindoro.

Additionally, the analysis showed that the use of an Artificial Neural Network (ANN) identified service quality, data privacy, and assurance as the main factors that impact service quality evaluation. These insights serve as valuable guidance for researchers and stakeholders interested in further studies in this domain, aiding in the development of comprehensive strategies to enhance the overall service quality and customer satisfaction. Thus, these findings substantially contribute to comprehending the essential elements in evaluating ISPs, leading to enhanced effectiveness, and encouraging ongoing use of this method. The developed models serve as informative guidelines for researchers and various sectors while offering valuable insights to ISPs to better understand and address the factors influencing consumer preferences in internet services.

6. Limitations and Future Research

This study analyzed customers’ preferences regarding various factors that affect the network’s performance. By analyzing the results, the researchers aimed to understand the main factors influencing internet users’ choices and degree of satisfaction. Despite the significant findings, several limitations were found. The fact that the participants were limited to residents of Occidental Mindoro, Luzon, the Philippines, is a significant limitation. Consequently, only 280 respondents from 10 municipalities comprised the study’s sample. Hence, this sample size may impact the study’s precision. Therefore, examining a larger sample that represents the various internet user perspectives would be beneficial to increase the study’s validity. Apart from the study’s limitations, the researchers could only examine eleven latent variables with the 81-item survey questions, limiting the data’s depth and the model’s standardization, potentially limiting the generalizability of findings. By adding more variables from various theories and models, it could be possible to improve the current study’s scope and reliability by thoroughly discussing its findings and results. The connections between the various latent variables with significant correlations could further support the study findings, making it possible for future researchers to provide more precise and accurate results.

In addition, the study recognized cultural factors’ significance in shaping consumer preferences for internet service providers (ISPs). While cultural differences impact perceptions of service quality, data privacy, and pricing, the research lacked an in-depth analysis of these influences. Therefore, investigating and examining cultural differences that affect perceptions of service quality, data privacy, customer satisfaction, and the impact of cultural diversity on the design and marketing of internet services can help ISPs better cater to the needs of various cultural groups.

Moreover, the researchers may have overlooked critical variables influencing customer loyalty and satisfaction, such as advertising effectiveness, availability of technical support services, and the impact of social media; thus, incorporating a comprehensive approach is necessary. Thus, using these methods, researchers can thoroughly study these variables’ impact on customer attitudes and behaviors, leading to a better understanding of the factors influencing customer satisfaction and loyalty in the ISP sector. Moreover, the researchers used SEM and an ANN to analyze complex relationships. However, it is essential to note that they may not have captured all cause-and-effect connections in the data. Thus, other researchers must carefully interpret the results and consider analyzing more data or adding variables to understand the dataset’s relationships better.