Shifting Perceptions and Behaviors: The Impact of Digitalization on Banking Services

Abstract

The rapid digitalization of banking services has transformed consumer interactions, necessitating a deeper understanding of the factors influencing online banking adoption. This research investigates the factors influencing consumer adoption in a country undergoing rapid digital transformation but still facing gaps in digital skills and infrastructure—Romania. The objective of the study is to analyze how key variables such as ease of use, security, speed, usefulness, and social pressure influence online banking behavior of Romanian consumers, especially the most digitally engaged ones. The study utilizes a multi-method empirical approach, hypothesis testing, binary logistic regression for prediction modeling, and segmentation analysis to offer a comprehensive view of customer behavior. The findings identify essential adoption drivers and separate customer profiles, providing useful information for financial organizations aiming to enhance their digital strategy. Perceived ease of use and perceived security are primary factors influencing adoption; nevertheless, decision tree analysis indicates that speed and usefulness have a more significant impact than logistic regression implies, but social pressure unexpectedly serves as an impediment. These results highlight the necessity for banks to customize their digital services, harmonizing security and user-friendliness with improved efficiency and usefulness to promote broader adoption in emerging digital economies like Romania.

1. Introduction

Due to digital transformation, the banking industry is experiencing substantial changes and diversification in its business operations. Furthermore, corporations, governments, and researchers all now place a high priority on the digitization of commercial banking activities. A substantial change in the banking sector has been brought about by the onset of digital transformation, which has redefined conventional banking procedures and transformed the way financial institutions engage with their clientele [1,2,3]. By implementing innovative technologies such as cloud computing, blockchain and artificial intelligence, banks are improving security, customer experience and streamlining operations [4]. This change improves both the efficiency and accessibility of financial services [5]. This stimulates creativity in financial services as it enables organizations to offer their customers customized solutions tailored to their changing needs. Consequently, the banking sector is constantly changing, and digitization will have a significant impact on how future customers will receive financial services.

Digitization of the banking sector is essential for consumers and businesses as it facilitates the entry of companies into the digital economy [6]. Banks are essential to help businesses by providing easy and accessible financial services to customers. These solutions improve the e-commerce experience and increase customer satisfaction and trust by processing payments on company websites in a secure and efficient way. In addition, digital banking improves operational efficiency and growth by automating transactions, streamlining financial management and providing access to real-time analytics [7].

The driving forces behind this study are the need for financial institutions to change their strategies to adapt to changes in consumer behavior and the increasing digitization of banking services. Using multiple analytical methods, this study aims to bridge the gap between theoretical adoption models and actual data-driven observations. Banks will be able to better personalize their digital offerings by understanding how the four main drivers of adoption interact: usefulness, speed, security and ease of use. In addition, understanding the paradoxical effects of social pressure allows rethinking user interaction and marketing strategies to increase adoption rates.

The study focuses on 19–35-year-olds, as they are the most financially evolved and technology-engaged group. Because they are digital natives, they have adopted online payments and banking faster than previous generations. In addition, they are significantly influencing the growth of e-commerce, encouraging cashless transactions and influencing financial trends.

This generation is particularly receptive to new banking technologies and digital literacy initiatives due to their current stage of transition from students managing their first bank accounts to professionals making important financial decisions. They are an important part of future financial progress because of their ability to evolve and grow.

The research focuses on this market and offers suggestions on how best to integrate digital banking and e-commerce with the next generation of customers.

Despite the fact that a lot of research has been conducted on the adoption of online banking, most of it has focused on specific aspects such as ease of use, usefulness, social pressure, speed, trust in artificial intelligence and adaptability. Moreover, mobile banking and internet banking usage were analyzed along with online purchasing decisions. Complex analytical methods such as segmentation analysis and predictive modeling are often not integrated to provide a deeper understanding of customer behavior. In addition, new machine learning techniques such as decision trees could help us to better understand the relative importance of different adoption factors. Conventional models such as logistic regression are still popular. Moreover, little is known about how social pressure influences adoption, in particular its surprising deterrent role.

2. Literature Review

Digital banking is essential for advancing e-commerce by facilitating seamless online transactions, safe digital payments, and enhancing financial literacy. Consequently, the involvement of banks in the community enhances financial inclusion through digital banking tools, thereby empowering citizens, particularly in underdeveloped regions, by granting access to online financial services [8]. Banks are investing in digital literacy initiatives to educate customers on digital finance, enabling individuals to acquire the skills required for confident participation in e-commerce and digital transactions [9]. The promotion of cashless transactions through digital wallets, mobile payments, and online banking has expedited the expansion of e-commerce, enhancing the speed and efficiency of transactions [10,11].

Building on the broader literature on digital banking, it is essential to analyze how customer attitudes have evolved in response to these technological changes.

2.1. Customer Attitudes Regarding Digitalization in the Banking Industry

The emergence of digital banking has transformed client expectations from conventional, branch-centric interactions to fluid, on-demand services accessible through several channels. Customers currently desire:

• Convenience—Mobile banking applications, AI-powered chatbots, and automated services provide round-the-clock access to banking without physical limitations [12,13,14,15,16,17].
• Personalization—Artificial intelligence and big data empower banks to provide customized financial solutions by anticipating customer demands and suggesting suitable products [12,13,16,18,19].
• Security and Trust—Given the escalating risks of cyber threats, banks must reconcile innovation with stringent security protocols to uphold customer trust. The reliance on the bank’s digital platform significantly influences ongoing utilization [20,21,22,23,24,25].

Studies continually emphasize trust and perceived usability as key determinants influencing the uptake of digital banking services [17,26,27,28], while factors influencing customer satisfaction encompass the quality of application functioning, user interface design, and the accessibility to customer assistance [29]. Millennials and Gen Z, as well as persons with advanced digital literacy, are generally more amenable to digital banking, valuing its rapidity and efficacy [3,30,31]. Conversely, Baby Boomers and Generation X frequently exhibit reluctance, citing apprehensions over security, unfamiliarity, and a predilection for face-to-face encounters [32,33]. In this context, Larsson and Viitaoja (2017) highlight that building customer loyalty in digital banking requires overcoming challenges related to digital CRM, as staff often struggle to replicate personalized interactions and relationship-building processes in online environments [34]. With the ongoing expansion of digital usage, users are placing greater importance on individualized and customized experiences.

Customers prioritize the service quality, usability, and transaction speed of digital banking, which substantially affect customer happiness, loyalty, and, consequently, the bank’s financial performance [29,35,36,37].

While attitudes capture perceptions and expectations, behaviors reveal how consumers actually engage with digital banking services.

2.2. Customer Behaviors Towards Digitalization in Banking

Customer behavior toward digital banking is shaped by psychological comfort, perceived security, ease of use, generational attitudes, and the staged adoption of services. Trust-building, personalization, and simplicity are key to increasing adoption and satisfaction.

According to the literature, customers adopt digital banking tools progressively, beginning with basic informational services such as checking account balances. As comfort and familiarity grow, they move on to more complex transactional services like money transfers. Many customers begin to explore non-bank digital financial services only after acquiring confidence in these technologies [38,39].

On the other hand, mobile banking has been the favored choice among younger people among the numerous digital platforms. These individuals are attracted to its convenience, time efficiency, and incentive-driven attributes; yet, security apprehensions persist as a prevalent obstacle to complete adoption [40]. The caliber of customer experience and perceived security are pivotal elements affecting ongoing usage and the propensity to endorse digital platforms.

Performance expectancy and social presence also play a key role. People are more inclined to adopt digital banking services when they believe the tools are reliable, efficient, and offer a sense of personal engagement [41,42]. This is especially crucial for fostering trust among various generations. Generations X and Y demonstrate differing degrees of confidence and familiarity with digital tools, resulting in distinct patterns of adoption and utilization.

Millennials and Gen Z, being younger demographics, exhibit more digital proficiency and frequently evaluate banking services. They assess online functionalities, customer service, prices, rewards, and loan alternatives, often changing banks in pursuit of superior digital experiences or more advantageous offerings [3,18,43]. A growing number of customers opt to have accounts with various banks to capitalize on specialized services, superior rates, or diminished fees.

In reaction to the evolving and competitive digital environment, banks are swiftly innovating to attract and keep clientele. The emergence of internet technologies has allowed institutions to provide more convenient, efficient, and tailored services. Attributes such online and mobile banking systems, AI-enhanced customer service, digital payment solutions, and customized financial offerings enable banks to distinguish themselves and deliver an exceptional client experience and maintain competitiveness in a swiftly changing market.

Having discussed global patterns of adoption, it is now important to narrow the focus to Romania, a country with rapid digital development but notable challenges.

2.3. Digitalization in Romania

The COVID-19 epidemic expedited the adoption of digital banking, resulting in increased reliance on online and mobile banking and a decline in branch visits [44,45,46]. There is an increasing demand for integrated, customized digital experiences, AI-enhanced customer assistance, and improved cybersecurity protocols. The use of cash has diminished in preference for contactless payments and electronic wallets, although hybrid banking models that integrate digital and physical services continue to be pertinent [47]. Customers have adopted FinTech technologies, open banking, and sustainable banking efforts [48]. As digital transformation progresses, banks must innovate and emphasize security, convenience, and personalization to satisfy changing client expectations.

As of December 2024, Romania has attained considerable digital penetration, with 94.57% of homes possessing internet connectivity, a notable rise from 42% in December 2010 [49]. This expansion is supported by a high-speed internet infrastructure, placing Romania among the leading countries in Europe for internet speeds. Notwithstanding these developments, difficulties endure, especially in rural regions where internet connectivity remains inferior to that of urban places.

According to the Eurostat Database (2023), merely 27.73% of adults in Romania possessed basic or above-basic digital skills, positioning the country near the bottom of the EU27 rankings. Bulgaria had a rate of 35.5%, whilst Poland attained 44.3%. In 2023, the Netherlands had the highest level at 83%, followed closely by Finland at 82%, while the EU27 average stood at 55.5%. Moreover, enterprises in Romania are similarly deficient in digital adoption. In 2023, 72% of Romanian enterprises, excluding those in the financial sector, agriculture, forestry, fishing, and mining, exhibited a markedly low digital intensity index [50]. This data underscores the considerable disparity in digital integration across the corporate sector, as numerous organizations continue to grapple with the use of modern digital technology and solutions.

Robust data indicates convergence in e-banking utilization among EU nations [51]. Although Romania has advanced in developing its digital infrastructure, additional initiatives are required to improve digital competencies and technology utilization among its citizens and enterprises to fully leverage the advantages of digitalization.

Although these insights are valuable and Romania provides a unique setting—with high internet penetration vs. low digital literacy—important gaps still persist, and this research is justified as a further investigation.

2.4. Research Gap

This study is driven by the growing digitalization of banking services and the necessity for financial institutions to enhance their strategies in response to changing consumer behaviors. This research seeks to reconcile theoretical adoption models with empirical, data-driven insights through the integration of several analytical methodologies. Comprehending the interaction among principal adoption factors—Perceived Ease of Use (PEU), Perceived Security (SEC), Perceived Speed (SPD), Perceived Usefulness (PU), and Perceived Social Pressure (SP)—will enable banks to customize their digital services more efficiently. Moreover, revealing the paradoxical impact of social pressure presents a chance to reevaluate marketing and user interaction techniques to enhance adoption rates.

The swift proliferation of digital banking and financial technologies has revolutionized customer engagement with banking services. Notwithstanding these gains, adoption rates varied markedly among various demographic and behavioral segments, underscoring the necessity to comprehend the principal drivers and obstacles affecting customer choices. Financial institutions seek to improve consumer engagement, trust, and security while tackling issues of usability, fraud prevention, and digital literacy. Furthermore, as AI-driven banking, cryptocurrencies, and open banking ecosystems proliferate, there is an increasing necessity to investigate how these technologies influence customer views and banking behaviors [52]. This project aims to address existing knowledge gaps and deliver data-driven insights to enhance digital banking strategy.

3. Methodology

The study employed a systematic analytical approach, commencing with hypothesis testing to investigate the correlations between customer perceptions and the uptake of online banking.

Although a considerable number of studies have been conducted on online banking adoption, the majority of studies emphasize technical elements, security concerns, or general consumer views, frequently overlooking the intricate interplay among behavioral, psychological, and technological components. The Technology Acceptance Model (TAM) has been extensively utilized; yet, its amalgamation with sophisticated analytical techniques—such as predictive modeling, segmentation analysis, and machine learning methodologies like decision trees—remains constrained. Conventional methods, such logistic regression, offer significant insights but may inadequately represent nonlinear relationships and threshold effects that impact adoption. Moreover, societal pressure has been little examined, especially its unforeseen function as a deterrent rather than a catalyst for adoption. Moreover, research frequently neglects underrepresented customer demographics, such as those with restricted digital literacy or financial access, resulting in deficiencies in comprehending how various segments interact with digital banking. This study addresses these gaps by utilizing a multi-method approach—integrating hypothesis testing, logistic regression, decision tree modeling, and cluster analysis—to offer a thorough and practical insight into the dynamics of online banking adoption in Romania, a topic that has been underexplored by researchers. Hypotheses were developed based on recognized technology adoption theories, particularly the Technology Acceptance Model, which emphasizes the significance of perceived ease of use (PEU) and perceived usefulness (PU) in technology adoption. Supplementary hypotheses examined variables including perceived security (SEC) [25,28,32,33], perceived social pressure (SP) [11,30,31], and Trust/Agreement with AI in Banking (AI) [14,18] to evaluate their impact on customers’ propensity to embrace digital banking services. In order to demonstrate the foundation of the hypotheses,

Table 1. Hypotheses development.

Hypothesis	Statement	Questionnaire Items Used in Testing the Hypothesis	Supporting References
H1	Perceived ease of use influences intention to use new technologies from the banking sector	Q18.2–Q18.5	[27,30,53,54,55,56,57,58,59]
H2	Perceived usefulness influences intention to use the new technologies from the banking sector	Q18.1–Q18.5	[30,53,54,55,56,57,58,59,60]
H3.1	The adoption of new banking technologies (internet banking) enhances customers’ online purchasing decisions	Q9.1–Q13	[9,60,61]
H3.2	The adoption of new banking technologies (mobile banking) enhances customers’ online purchasing decisions	Q9.2–Q13	[31,32,35,37,39,40]
H4	Consumers who perceive online banking services as secure are more likely to use them frequently	Q9.1–Q16.1/Q9.2–Q16.1	[28,30,60]
H5	Consumers who perceive online (internet and mobile) banking as well-promoted are more likely to be well-acquainted with online banking services	Q8.1, Q8.2–Q17	[2,3,8,9,51,52]
H6	Higher income consumers are more likely to use online banking services for financial transactions	Q15	[8,13,27,38,51]
H7	Consumers who frequently shop online are more likely to perform future banking transactions online	Q18.1–Q18.5	[9,31,32]
H8	Consumers who currently use mobile banking are more likely to intend to continue using digital banking services in the future	Q9.2–Q18.5	[35,40,42,47,57,58]
H9	Perceived social pressure influences the decision to use online banking services	Q16.3–Q9.1, Q9.2	[11,32,39,41,42]
H10	Trust in artificial intelligence in banking services positively impacts the intention to perform financial transactions online	Q16.7–Q18.5	[12,14,18,19,20,30,61]
H11	Consumers who find online banking services faster and more convenient are more likely to use them	Q18.3–Q9.2	[2,3,35,57,58,59,60]
H12	The level of agreement with the statement that ‘adaptation to online technologies is necessary’ is positively correlated with the frequency of online banking use	Q16.2–Q9.1, Q9.2	[2,7,21,39]

presents the existing knowledge and theories. Non-parametric tests, including Spearman’s correlation and Kruskal–Wallis tests, were employed because of the non-normal distribution of variables, hence ensuring robustness in statistical inference.

Based on the hypothesis testing results, binary logistic regression was conducted to measure the prediction strength of significant behavioral and demographic variables on online banking adoption. This analysis determined the primary factors influencing adoption while accounting for other possible effects. The results provide insights into the likelihood of adoption based on various consumer characteristics, laying the groundwork for additional predictive modeling.

A Classification and Regression Tree analysis was performed to identify nonlinear correlations and intricate decision patterns. Decision trees provide a comprehensible, hierarchical framework for elucidating the interactions among various elements that affect adoption [62,63]. In contrast to logistic regression, which presumes linear connections, decision trees facilitate threshold effects and conditional interactions, rendering them especially effective for discerning key decision-making factors in online banking adoption [64].

After identifying key predictors, a K-Means clustering analysis was conducted to categorize customers into separate groups based on their beliefs and behaviors [65,66]. Clustering facilitated the identification of inherent segments within the consumer base, distinguishing between technologically proficient users, moderate adopters, skeptical individuals, and faithful traditional banking customers. This segmentation offers practical information for financial institutions to customize their marketing tactics, improve digital banking experiences, and tackle specific obstacles to adoption among reluctant consumers.

The comprehensive analytical procedure was executed utilizing IBM SPSS Statistics V26, which offered a formidable platform for data preparation, hypothesis testing, regression modeling, decision tree development, and clustering. This sequential and data-driven methodology guarantees that each analytical phase is informed by the preceding findings, culminating in a thorough and organized comprehension of consumer adoption behavior.

This methodical approach guarantees that the research outcomes are both statistically robust and pragmatically relevant, providing evidence-based recommendations for banks to enhance their digital banking services and elevate user engagement across various customer categories.

Design of Questionnaires

A systematic questionnaire was administered utilizing the CAWI (Computer-Assisted Web Interviewing) method to examine consumer behavior about banking digitization from October to December 2024. This digital method guaranteed extensive accessibility and convenience for participants. The questionnaire comprised demographic and behavioral items, including closed-ended questions (single choice and Likert-scale formats) to guarantee uniform and analyzable replies. The poll encompassed essential characteristics including age, gender, domicile, educational attainment, and income, succeeded by targeted inquiries regarding the utilization of digital banking services, trust in digital platforms, and perceptions of technology in banking.

Table 2. Measurement Mapping Table for Questionnaire Constructs.

Construct (Abbr.)	Questionnaire Item	Example Item/Variable Label Used in Analysis	Scale & Coding	Used in H#/Analyses
EDU, AGE	EDU, AGE	As collected	Ordinal/numeric	Logistic; CART
INC	INC	Monthly income	Ordinal bands	H6 (KW); Logistic; CART
familiarity online banking	Q8.1/Q8.2	What types of electronic banking services are you familiar with? (internet, mobile banking)	Likert (1–5)	H5 (KW)
MB_USE	Q9.1	Frequency of mobile banking use	4-level ordinal (or binary for H8)	H3.2 (KW), H8 (U test)
IB_USE	Q9.2	Frequency of internet banking use	4-level ordinal (or binary for H8)	DV in Logistic/CART; also H3.1 (KW)
OPD	Q13	“New banking technologies impact my online purchasing decisions”	Likert (1–5)	DV in H3.1–H3.2
online card use	Q15	“Do you use your card online”	Categorical	H6 (KW)
SEC	Q16.1	“Online transactions are secure”	Likert (1–5)	H4; Spearman; Logistic; CART
ADAPT	Q16.2	“Adapting to online technologies is necessary”	Likert (1–5)	H12; Spearman; Logistic; CART
SP	Q16.3	“People important to me think I should use online banking”	Likert (1–5)	H9; Mann–Whitney; Logistic; CART
AI	Q16.7	“Using AI in banking is acceptable/I trust AI features”	Likert (1–5)	H10; Spearman; Logistic; CART
online banking promotion	Q17	“To what extent do you believe that Internet Banking services are promoted by banks so that they are well-known to the public”	Likert (1–5)	H5 (KW)
PU	Q18.1	“Internet banking is useful”	Likert (1–5)	H2; Spearman; Logistic; CART
PEU	Q18.2	“Internet banking is easy to use”	Likert (1–5)	H1; Spearman; Logistic; CART
SPD	Q18.3	“Internet banking is fast/time-saving”	Likert (1–5)	H11; Spearman; Logistic; CART
use intent	Q18.5	“I will perform more transactions through online services”	Likert (1–5)	H1; Spearman

Source: author’s own elaboration.

maps each questionnaire construct to a representative survey item. It presents the measurement scale or coding and the analyses or hypotheses where it was applied.

Subsequent to the formulation of the questionnaire, a pre-test was executed on a limited sample to evaluate the clarity, logical coherence, and comprehensibility of the questions, as well as to pinpoint any challenges respondents may face in submitting answers. The pre-test comprised a sample of 10 participants, intended to identify and rectify potential flaws or ambiguities in phrasing. Following the outcomes of this pre-testing phase, numerous questions were amended for enhanced clarity, and supplementary response alternatives were incorporated as needed.

Romania was chosen as the geographic scope due to the academic affiliation of the research team as well as the overarching objective of examining the adoption of digital banking in an EU member state that is rapidly advancing its digital economy.

The respondents were recruited through digital channels, including online forums, university mailing list and social media platforms. Although the study used convenience sampling, the research focuses on individuals aged between 19 and 39 which are the most digitally engaged with the banking products and services.

A total of 481 surveys were gathered online and were authenticated. To ensure data quality and coherence with the target population, responses from individuals aged 40 and above were excluded during the data validation phase.

The age restriction was established to concentrate the research on young persons from Generation Y, also referred to as millennials—those born between 1980 and the early 2000s. The decision to restrict the age range of respondents was predicated on the idea that young millennials, aged 19 to 39, are the most engaged consumers of emerging financial technologies. They are deemed to symbolize the future and are therefore significantly pertinent to the subject.

Although the study employed a convenience sampling strategy, the achieved sample size of n = 481 respondents aged 19–39 can be considered reasonably representative of the target population of approximately 5.7 million Romanians (about 30% of Romania’s population) in this age group. Using a 95% confidence level and the most conservative assumption for proportions (p = 0.5), the margin of error for this sample is approximately ±4.5%. This indicates that the survey estimates can be generalized to the broader digitally active young adult population in Romania with acceptable precision, while acknowledging the inherent limitations of non-probability sampling.

4. Results

To address the identified research gaps, this study employs a multi-method approach that incorporates hypothesis testing, regression modeling, decision trees, and clustering. Hypotheses testing looks at the relationship between variables, logistic regression tests direct linear effects and decision trees capture non-linear, conditional, and threshold effects.

Before presenting the specific analyses, it is useful to outline how hypotheses were derived and tested. Based on the Technology Acceptance Model (TAM), one of the most widely used theoretical frameworks for explaining and predicting user acceptance of new technologies, the hypotheses were constructed and organized into several sections. These sections emphasize perceptions of using online banking services, consumer behaviors, intentions to adopt such services, and various factors influencing rapid adoption, such as trust in AI, perceived speed, and adaptability.

4.1. Hypothesis Testing Results

Normality tests were conducted using the Kolmogorov–Smirnov and Shapiro–Wilk tests, both of which confirmed significant deviations from normality across all ordinal and scale variables (p < 0.001). This finding was consistent across measures of banking familiarity, usage, perceived importance, and attitudinal agreement, indicating non-normal distributions throughout the dataset.

Given these results, parametric statistical methods—such as t-tests, ANOVA, and Pearson’s correlation—are not suitable, as they assume normality and may yield biased results. Instead, non-parametric alternatives were used: for correlation analyses, Spearman’s rank correlation, group comparisons Mann–Whitney U tests (for two groups) and Kruskal–Wallis tests (for multiple groups).

4.1.1. Perceptions of Online Banking Services

The first set of hypotheses examines how perceptions of ease of use and usefulness affect consumers’ intentions to adopt digital banking. Moreover, the research emphasis if the adoption of digital banking enhances customers’ online purchasing decision.

Hypothesis 1. (H1)

Perceived ease of use influences intention to use new technologies from the banking sector.

To examine if the perceived ease of use influences intention to use new technologies from the banking sector, Spearman’s rank-order correlation was conducted to assess the relationship between agreement on PEU of Internet Banking and agreement on future transactions using Internet Banking.

The results indicate a strong, positive correlation between the two variables, ρ(481) = 0.638, p < 0.001, suggesting that individuals who perceive internet banking as easy to use are more likely to intend to use banking technologies in the future [58,59,66]. The statistical significance at the 0.01 level (two-tailed) confirms the robustness of this relationship.

Hypothesis 2. (H2)

Perceived usefulness influences Intention to Use the new technologies from the banking sector.

To evaluate H2, Spearman’s rank-order correlation was conducted to examine the relationship between agreement on Usefulness of Internet Banking and agreement on Future Transactions using Internet Banking.

The results indicate a strong, positive correlation, ρ(481) = 0.631, p < 0.001, suggesting that individuals who perceive internet banking as useful (PU) are more likely to express an intention to use such technologies in the future [57,59,60,67]. The statistical significance at the 0.01 level (two-tailed) further supports the robustness of this relationship. Jena (2023) shows that non-users (especially senior citizens) have a significantly lower perception of technology’s usefulness and influence on financial decisions compared to occasional or regular users [46].

Hypothesis 3.1. (H3.1)

The adoption of new banking technologies (internet banking) enhances customers’ online purchasing decisions.

To assess if the adoption of internet banking enhances customers’ online purchasing decisions, an Independent-Samples Kruskal–Wallis Test was conducted to compare the impact of new technologies on buying decision across different levels of internet banking usage.

The Kruskal–Wallis test yielded a statistically significant result, H(3) = 20.599, p < 0.001, indicating that the distribution of Impact of New Technologies on Buying Decision differs significantly across categories of Internet Banking Usage. Post hoc pairwise comparisons with Bonferroni correction showed that respondents who do not use Internet Banking at all perceive a significantly lower impact of new technologies on their buying decisions compared to those who use internet banking sometimes (p = 0.013), often (p = 0.008), or very often (p < 0.001). However, no significant differences were observed among the sometimes, often, and very often user groups, suggesting that increased frequency of Internet Banking usage does not necessarily lead to a stronger perceived impact on OPD beyond a certain threshold. Kumar & Anand, 2016 emphasized that those unfamiliar with internet banking tend to view it as complex or intimidating, thus limiting their intention to use digital tools even in online shopping contexts [68]. Recent bibliometric analyses support these behavioral distinctions. For instance, Windasari et al. (2022), in their study on digital-only banking experiences, highlight how younger consumers (Gen Y and Z) perceive more value in digital financial tools, particularly when they have previous digital exposure [3]. In contrast, digitally inactive users remain skeptical about the relevance of such technologies to their daily choices [3].

Hypothesis 3.2. (H3.2)

The adoption of new banking technologies (mobile banking) enhances customers’ online purchasing decisions.

To examine if the adoption of mobile banking enhances customers’ online purchasing decisions, an Independent-Samples Kruskal–Wallis Test was conducted to compare the Impact of New Technologies on Buying Decision across different levels of Mobile Banking Usage.

The Kruskal–Wallis test revealed a statistically significant difference in perceived impact among different levels of mobile banking usage, H(3) = 25.227, p < 0.001, suggesting that individuals with varying usage patterns perceive the influence of banking technologies on purchasing decisions differently. Post hoc pairwise comparisons with Bonferroni correction revealed that respondents who do not use mobile banking at all report significantly lower perceived impact of new technologies on online purchases compared to those who use mobile banking often (p = 0.004) and very often (p < 0.001). However, no significant differences were found between sometimes and often users, nor between often and very often users, indicating that the influence of mobile banking on purchasing decisions (OPD) stabilizes beyond a certain usage threshold. Many researchers empirically demonstrate that e-payment methods (linked to mobile banking) directly impact online shopping behaviors and contribute to sales growth [11,27,69].

4.1.2. Consumer Behavior Regarding Online Banking

Beyond perceptions, it is also necessary to assess how security, promotion, and income shape actual consumer behaviors towards online banking services. Therefore, the following hypothesis were tested.

Hypothesis 4. (H4)

Consumers who perceive online banking services as secure are more likely to use them frequently.

To test H4, Spearman’s rank-order correlation was conducted to examine the relationships between Agreement on Security in Transactions, Usage of Internet Banking, and Usage of Mobile Banking.

A moderate, positive correlation was found between SEC in transactions and B_USE, ρ(481) = 0.400, p < 0.001, indicating that individuals who perceive online banking as secure tend to use Internet Banking more frequently [25,28,33].

A similar moderate, positive correlation was observed between SEC and MB_USE, ρ(481) = 0.377, p < 0.001, suggesting that security perceptions also influence mobile banking adoption [32].

Hypothesis 5. (H5)

Consumers who perceive online (internet and mobile) banking as well-promoted are more likely to be well-acquainted with online banking services.

To evaluate H5, an Independent-Samples Kruskal–Wallis Test was conducted to compare levels of Familiarity with Internet Banking and Mobile Banking across different Internet Banking Promotion categories.

The Kruskal–Wallis test for Familiarity with Internet Banking yielded a statistically significant result, H(4) = 33.266, p < 0.001, indicating that perceived promotion levels significantly influence how well consumers are acquainted with Internet Banking [3,11]. Similarly, for Familiarity with Mobile Banking, the Kruskal–Wallis test also returned a significant result, H(4) = 19.886, p = 0.001, suggesting that promotion efforts impact mobile banking familiarity as well [31,35].

Post hoc pairwise comparisons with Bonferroni correction revealed that respondents who perceive Internet Banking as “Very Well Promoted” exhibit significantly higher familiarity with Internet Banking compared to those who believe it is “Not Promoted Enough” (p < 0.003) and “Well promoted (p = 0.013). Interestingly, respondents that chose “Neutral” showed significantly lower levels of familiarity (p< 0.001) for both mobile and internet banking as compared with respondents who considered them as “Very Well Promoted”. There were no significant differences in familiarity levels between the other pairs.

Hypothesis 6. (H6)

Higher income consumers are more likely to use online banking services for financial transactions.

An Independent-Samples Kruskal–Wallis assess was performed to evaluate monthly income (INC) across varying amounts of Online Card Usage for Purchases in order to assess H6.

The Kruskal–Wallis test produced a statistically significant outcome, H(3) = 79.418, p < 0.001, demonstrating that income levels vary significantly according to consumers’ frequency of utilizing online banking services for transactions. Post hoc pairwise comparisons utilizing Bonferroni correction indicated that consumers who “Always” utilize online cards for purchases own considerably higher income levels than all other categories (p < 0.001 for all comparisons). Individuals who utilize online cards “Sometimes” exhibit considerably higher income levels compared to those who “Rarely” use them (p = 0.002) and show minor significance when compared to those who “Never” use them (p = 0.051). No substantial disparity in income was noted between individuals who utilize online cards “Rarely” and those who “Never” employ them (p = 1.000), indicating that the lowest-income consumers may entirely eschew online transactions. Additionally, additional studies indicate that higher income groups exhibit markedly stronger interaction with digital and online banking channels for ordinary financial activities [27,40].

4.1.3. Intention to Use Online Banking Services

After understanding perceptions and behaviors, evaluating consumers’ future intentions to use digital banking are important and indicate long-term adoption trends. Hence, H7 and H8 are tested.

Hypothesis 7. (H7)

Consumers who frequently shop online are more likely to perform future banking transactions online.

Spearman’s rank-order correlation was performed to assess the association between Agreement on the Usefulness of Internet Banking, likely due to frequent online shopping, and Agreement on Future Transactions utilizing Internet Banking for the evaluation of H7.

The findings reveal a robust positive association, ρ(481) = 0.631, p < 0.001, indicating that persons who regard Internet banking as beneficial—presumably due to regular online shopping—are more predisposed to engage in future online banking transactions [11,31]. The statistical significance at the 0.01 level (two-tailed) validates the strength of this link.

Hypothesis 8. (H8)

Consumers who currently use mobile banking are more likely to intend to continue using digital banking services in the future.

A Mann–Whitney U-Test was conducted to assess H8 by comparing future digital banking intentions between consumers who frequently utilize mobile banking and those who seldom or never engage with it.

The test revealed a statistically significant disparity in future banking intentions between regular mobile banking users (N = 319, Mean Rank = 276.12) and non-users or infrequent users (N = 162, Mean Rank = 171.85), U = 37,041.500, Z = 8.455, p < 0.001. The results offer robust empirical evidence for H8, demonstrating that present mobile banking users possess a markedly greater propensity to persist in digital financial activities [31,55]. The significant mean rank disparity between the two user groups highlights the critical role of initial adoption in promoting sustained engagement with digital banking services.

Hypothesis 9. (H9)

Perceived social pressure influences the decision to use online banking services.

To examine whether perceived social pressure (SP) influences the adoption of online banking, a Mann–Whitney U test was conducted, comparing agreement levels on social pressure between users and non-users of internet banking. The study of Ikhsan et al. 2025 [30] and Raj et al. incorporates social influence as a significant predictor of user intention to adopt AI-enabled banking, suggesting that perceived pressure from peers and society affects digital banking decisions [11].

The results indicated no significant difference in perceived social pressure between users (N = 390, Mean Rank = 244.58) and non-users (N = 91, Mean rank = 225.68) of online banking, U = 19,139.50, z = 1.238, p = 0.216. The effect size was small, suggesting that social pressure does not play a significant role in distinguishing between adopters and non-adopters of online banking.

Still, when comparing agreement on social pressure between consumers who regularly use mobile banking and those who never or rarely use it, the Mann–Whitney U test shows a statistically significant difference in social pressure perceptions between the two groups, U = 29,430.500, Z = 2.642, p = 0.008. Regular mobile banking users (N = 319, Mean Rank = 252.26) reported significantly higher levels of agreement on SP compared to non-users or rare users (N = 162, Mean Rank = 218.83).

4.1.4. Factors Influencing the Use of Online Banking Services

Finally, the influence of trust in artificial intelligence (AI), perceived speed (SPD), and adaptability (ADAPT) is assessed, broadening the scope of adoption factors.

Hypothesis 10. (H10)

Trust in artificial intelligence in banking services positively impacts the intention to perform financial transactions online.

Spearman’s rank-order correlation was performed to examine H10, examining the association between Agreement on AI Use in Banking and Agreement on Future Transactions via Internet Banking.

A moderate, positive connection exists between faith in AI and the intention to utilize online banking services, ρ(481) = 0.349, p < 0.001. The statistical significance at the 0.01 level (two-tailed) validates the strength of this link.

The findings substantiate H10, demonstrating that increased trust in AI-driven banking services correlates with a heightened intention to engage in future online financial transactions. Confidence in AI voice assistants significantly influences the adoption of digital banking among Generation Z, particularly for transactional interactions [14,18].

Hypothesis 11. (H11)

Consumers who find online banking services faster and more convenient are more likely to use them.

Spearman’s rank-order correlation was performed to assess H11, investigating the association among Agreement on Speed of Internet Banking, Usage of Internet Banking, and Usage of Mobile Banking.

A moderate, positive connection was identified between Agreement on Speed of Internet Banking (SPD) and Usage of Internet Banking (IB_USE), ρ(481) = 0.434, p < 0.001, indicating that consumers who regard internet banking as speedier are more inclined to utilize it [39,40,54,55]. A comparable moderate connection was seen between Agreement on Speed of Internet Banking (SPD) and Usage of Mobile Banking(MB_USE), ρ(481) = 0.355, p < 0.001, suggesting that perceived convenience additionally affects mobile banking uptake.

Hypothesis 12. (H12)

The level of agreement with the statement that “adaptation to online technologies is necessary” is positively correlated with the frequency of online banking use.

Spearman’s rank-order correlation was performed to assess H12, analyzing the correlations among Agreement on the Necessity of Rapid Adaptation to Online Technologies, Internet Banking Usage, and Mobile Banking Usage.

A weak-to-moderate positive association exists between Agreement on Necessity of Adaptation (ADAPT) and Usage of Internet Banking (IB_USE), ρ(481) = 0.250, p < 0.001, indicating that those who see the necessity of adapting to online technologies are more likely to utilize Internet Banking often [3,27,46]. A slightly stronger correlation was observed between Agreement on Necessity of Adaptation (ADAPT) and Usage of Mobile Banking (MB_USE), ρ(481) = 0.277, p < 0.001, indicating that perceived necessity is more strongly associated with mobile banking adoption.

4.2. Predictive Modeling of Consumer Adoption of Internet Banking Services

To complement hypothesis testing, predictive modeling through logistic regression quantifies the likelihood of adoption based on consumer characteristics. A binary logistic regression analysis was conducted to examine the factors predicting consumer adoption of online banking services. The dependent variable was the utilization of online banking (0 = Not utilizing IB, 1 = Utilizing IB). The independent factors comprised PEU, PU, SEC, AI, SP, SPD, ADAPT, INC, EDU, AGE.

The logistic regression model was statistically significant, χ²(10) = 142.474, p < 0.001, demonstrating that the factors collectively elucidated online banking adoption. The model explained 41.3% of the variance in banking adoption (Nagelkerke R² = 0.413) and accurately identified 85.7% of cases. The Hosmer–Lemeshow test (χ²(8) = 4.327, p = 0.827) indicated that the model adequately fit the data.

PEU was identified as the most significant predictor of online banking adoption (B = 1.041, SE = 0.211, Wald = 24.324, p < 0.001). The odds ratio (Exp(B) = 2.832) indicates that consumers who view internet banking as user-friendly are 183.2% more inclined to utilize these services. This discovery corroborates H1 and reinforces the TAM framework, highlighting that usability is a crucial factor in technological uptake.

Consumers’ agreement with the statement that online banking provides secure transactions significantly predicted adoption (B = 0.617, SE = 0.154, Wald = 16.045, p < 0.001). The odds ratio (Exp(B) = 1.853) suggests that a heightened SEC increases the likelihood of online banking adoption by 85.3%. These data support H4 and emphasize that security concerns are a key factor in the adoption of digital banking services, highlighting the need to implement fraud protection measures and consumer confidence building strategies.

Surprisingly, SP was found to negatively affect the adoption of online banking (B = −0.525, SE = 0.181, Wald = 8.392, p = 0.004). The odds ratio (Exp(B) = 0.592) suggests that individuals who experience more intense social pressure to use online banking are 41% less likely to adopt it. This result contradicts H9, which posited that there is no effect, and suggests that external pressure may engender resistance rather than motivation for adoption. A possible explanation is that individuals who feel coerced into using online banking may experience psychological reactance, leading them to reject or delay adoption.

Although not statistically significant at the conventional threshold, monthly INC approached significance as a predictor (B = 0.339, SE = 0.177, Wald = 3.695, p = 0.055 The odds ratio (Exp(B) = 1.404) indicates that individuals with higher incomes are 40.4% more likely to use online banking, thus supporting H6. This finding is consistent with previous studies suggesting that individuals with higher financial resources benefit from increased access to digital banking infrastructure.

Several variables were not significant predictors of online banking adoption, including PU (H2), AI (H10), SPD (H11), ADAPT (H12), education level and age. While these factors may indirectly influence adoption, they did not demonstrate a direct effect in this model.

The estimated logistic regression equation is:

$$log\left({\displaystyle \frac{P\left(Using\right)}{1-P\left(Using\right)}}\right)=-2.319+\left(1.041\times EaseofUse\right)+\left(0.617\times Security\right)-\left(0.525\times SocialPressure\right)+\left(0.339\times Income\right)$$

where P(Using) represents the probability of adopting online banking.

The findings indicate that perceived ease of use and security perception are the most important factors driving online banking adoption, whereas social pressure unexpectedly discourages adoption. that PEU and SEC are key drivers. However, the decision tree identified SPD and PU as more influential than logistic regression suggested.

4.3. Classification and Regression Tree Analysis

Since regression captures only linear relationships, decision tree analysis is employed to uncover nonlinear patterns and threshold effects in consumer decisions. A Classification and Regression Tree analysis was performed to determine the primary predictors of online banking uptake (Figure 1). The decision tree model was constructed with the Classification and Regression Trees technique, employing Gini impurity as the criterion for splitting. The model employed a maximum depth of five levels, requiring a minimum of 20 instances per parent node and 10 cases per child node. The dependent variable was the IB_USE (0 = Not utilized, 1 = Utilized). The independent factors comprised PEU, PU, SEC, AI, SP, SPT, ADAPT, INC, EDU, and AGE. The decision tree analysis sought to offer a clear, hierarchical ranking of the elements affecting online banking adoption, facilitating the identification of principal determinants and threshold effects in consumer decision-making.

The decision tree model achieved a classification accuracy of 86.7%, demonstrating robust predictive capability. The algorithm accurately classified 94.4% of internet banking users, but its accuracy for non-users was lower at 53.8%. The risk estimate (error rate) was 0.133 (13.3%), indicating that the model misclassified roughly 13% of cases. The results demonstrate that the model effectively identifies online banking customers; however, it exhibits reduced reliability in predicting non-adopters of the service.

The Variable Importance table (

Table 3. Variable Importance table.

Predictor	Importance Score	Normalized Importance
Agreement on Perceived Ease of Use of Internet Banking	0.084	100.00%
Agreement on Speed of Internet Banking	0.047	56.60%
Agreement on Usefulness of Internet Banking	0.042	50.70%
Agreement on Security in Transactions	0.032	38.00%
Agreement on Social Pressure	0.014	16.30%
Agreement on Necessity of Digital Adaptation	0.012	14.50%
Agreement on success of AI implementation	0.011	13.60%
Monthly Income	0.009	11.20%
Age	0.006	6.70%
Education Level	0.005	6.20%

Source: author’s own elaboration.

) ranked predictors based on their relative influence in classifying consumers into adopters and non-adopters of online banking.

The perceived simplicity of use emerged as the paramount predictor of online banking adoption, attaining a significance score of 100%. This discovery corroborates the findings of H1, validating the idea that intuitive, user-friendly interfaces enhance adoption rates. The decision tree structure corroborated that simplicity of use was the initial splitting criterion, signifying its role as the principal determinant in customers’ decision-making.

The perceived speed of online banking services was ranked as the second most significant aspect, with an importance of 56.6%. This finding indicates that customers prioritize the efficiency of digital banking and favor expedited transactions when determining the adoption of these services. This outcome, which lacked statistical significance in logistic regression, underscores the efficacy of tree-based methodologies in identifying nonlinear effects and threshold-driven decision-making.

The perceived utility of online banking was ranked third, with an importance of 50.7%, indicating that customers evaluate the practical advantages of online banking in their adoption decisions. Although usefulness was not statistically significant in logistic regression (p = 0.099), its relevance in the decision tree indicates that it may influence classification, potentially interacting with other variables like simplicity of use.

Security concerns were identified as the fourth most critical component (38.0% relevance), aligning with the logistic regression analysis, in which it emerged as a statistically significant predictor (Exp(B) = 1.85, p < 0.001). This research corroborates H4 and verifies that consumers who regard online banking as secure are more inclined to adopt it.

Social Pressure (16.3%): Although social pressure was identified as a predictor, its very modest significance indicates that external influence does not significantly affect adoption decisions. This conclusion corresponds with the logistic regression outcome, wherein social pressure exerted a negative influence (Exp(B) = 0.592, p = 0.004), suggesting that perceived coercion may result in resistance instead of acquiescence.

The necessity of digital adaptation (14.5%) and the successful implementation of AI (13.6%) contributed minimally to the model, suggesting that although consumers recognize the significance of digital banking adaptation and trust in AI-based banking services, these factors are not central to their decision-making process.

Income (11.2%), Age (6.7%), and Educational Attainment (6.2%): Demographic characteristics exhibited limited predictive capability, indicating that attitudes toward technology (ease of use, security, and speed) are far more impactful than socioeconomic position.

Both theories concur that usability and security are fundamental determinants. Nonetheless, the decision tree indicated that Speed and Usefulness were more influential than the logistic regression proposed. This underscores the importance of integrating several modeling methodologies to more accurately represent consumer decision-making.

4.4. Consumer Segmentation and Cluster Profile Development

Finally, segmentation analysis groups consumers into distinct clusters, offering practical insights for tailoring banking strategies.

A K-Means clustering analysis was performed to categorize consumers according to their attitudes and behaviors about online banking adoption. The clustering model comprised seven essential factors derived from the logistic regression and decision tree analyses: PEU, PU, SPD, SEC, SP, INC, IB_USE.

The K-Means clustering algorithm was performed with four clusters (k = 4), determined based on interpretability and differentiation among consumer groups. The standardized (Z-score) values of the selected variables were used to ensure all predictors contributed equally to the clustering solution. The algorithm converged after 19 iterations, indicating stability in the solution.

The ANOVA results indicated that all seven clustering variables significantly differentiated the clusters (p < 0.001 for all predictors), confirming that the segmentation approach effectively captured meaningful distinctions in consumer behavior. The final cluster sizes were as follows:

• Cluster 1: 132 consumers (27% of the sample)
• Cluster 2: 231 consumers (48% of the sample)
• Cluster 3: 65 consumers (14% of the sample)
• Cluster 4: 53 consumers (11% of the sample)

The F-values for key predictors ranged from 22.43 (income level) to 666.98 (online banking usage), with the highest values observed for PEU, SPD, and IB_USE, confirming their central role in differentiating consumer segments.

The Final Cluster Centers Table provided insights into the distinct characteristics of each consumer segment (

Table 4. Final Cluster Centers Table.

Variables	Cluster
	1	2	3	4
Agreement on PEU of Internet Banking	0.0993	0.5700	−0.6384	−1.9487
Agreement on SPD of Internet Banking	0.1029	0.5200	−0.2427	−2.2252
Agreement on PU of Internet Banking	0.0806	0.5243	−0.3198	−2.0938
Agreement on SEC in Transactions	−0.2957	0.5663	−0.5388	−1.0709
Agreement on SP	−0.7118	0.5608	0.1287	−0.8295
Monthly Income	−0.3095	0.3637	−0.4164	−0.3035
Usage of internet banking (binary)	0.4825	0.4715	−2.0680	−0.7206

Source: author’s own elaboration.

).

Cluster 1: Moderate Digital Banking Users (N = 132, 27%): This category comprises prudent digital banking adopters who perceive online banking as beneficial and convenient, however harbor modest apprehensions regarding security. They express marginally favorable views of PEU, SPD, and PU (~0.1), however maintain a neutral to slightly adverse perception of SEC(−0.29). Their influence from SP is below average (−0.71), indicating that external factors have minimal impact on their decision-making. Furthermore, they belong to a lower-income category (−0.30), which may influence their interaction with financial services. Notwithstanding these apprehensions, this demographic is still very inclined to utilize online banking (0.48). These consumers may gain from security-oriented messaging and confidence-enhancing strategies that mitigate their trust issues while emphasizing the usability and convenience of digital banking.

Cluster 2: Technologically proficient, highly engaged users (N = 231, 48%): The majority segment, representing 48% of the sample, is made up of highly engaged, technology enthusiastic individuals who actively use online banking services. They demonstrate superior ratings of PEU, SPD, PU (~0.52–0.57) and indicate the highest level of SEC (0.57) of all groups. In contrast to other groups, they show low sensitivity to SP (0.56), indicating that peer influence and social norms may facilitate their persistent adoption. Due to a higher average income (0.36), financial barriers do not hinder their access to digital banking. Their predilection for using online banking (IB_USE) is considerable (0.47), placing them as an ideal demographic for advanced digital services, including financial tools optimized via artificial intelligence and personalized banking experiences. Targeting this demographic via advanced banking functionalities, automation tools and loyalty programs can greatly increase their engagement.

Cluster 3: Skeptical Digital Users (N = 65.14%): This segment comprises reluctant and skeptical consumers who demonstrate poor confidence in internet banking. They indicate subpar evaluations of PEU (−0.64), SPD (−0.24), and PU (−0.32), suggesting challenges in accessing digital banking services. Moreover, their impression of security is significantly poor (SEC = -0.54), which reinforces their hesitance to utilize digital financial services. In contrast to other groups, they indicate a neutral level of SP (0.13), signifying that external pressures neither promote nor inhibit their use of internet banking. Their income levels are subpar (INC = −0.42), maybe leading to their disengagement from financial technologies. This group, with a minimal probability of utilizing internet banking (IB_USE = −2.07), constitutes one of the least active customer sectors. They may benefit from educational initiatives, improved user interface designs, and security assurance programs to build confidence and encourage adoption.

Cluster 4: Hesitant traditionalists (N = 53.11%): This small proportion (11% of the sample) is made up of traditional banking devotees who strongly oppose the introduction of digital banking. They show clearly negative ratings in terms of PEU (−1.95), SPD (−2.22) and PU (−2.09), making them the least likely to recognize the value of online banking. Their perceptions of security are significantly negative (SEC = −1.07), reinforcing their resilience. In contrast to other undecided groups, they exhibit a moderately negative influence of SP (−0.83), highlighting their deliberate resistance to peer or societal encouragement to adopt digital banking. Their income is marginally below average (INC = −0.30), which may restrict their interaction with financial technology. The lower propensity of this group for online banking (−0.72) suggests that they are unlikely to adopt digital banking without considerable external incentives. To facilitate acceptance by reluctant consumers, strategies including gradual digital integration, financial incentives and hybrid banking models can be implemented.

The four identified consumer segments provide a structured understanding of online banking adoption behaviors. While tech-savvy users (Cluster 2) are highly engaged, moderate adopters (Cluster 1) have security concerns but remain active users. In contrast, skeptical users (Cluster 3) require targeted education and usability improvements, while reluctant traditionalists (Cluster 4) exhibit strong resistance to digital banking. These insights can help financial institutions develop targeted strategies to enhance engagement, address barriers, and optimize digital banking adoption across diverse consumer segments.

5. Discussion

The methodology, which includes hypothesis testing, logistic regression, decision tree modeling, and cluster analysis, offers a thorough understanding of online banking usage. Throughout all methodologies, user-friendliness proved to be the predominant element. Regression established its statistical significance, decision trees identified it as key to adoption routes, and clustering underscored its relevance among technologically adept users, although less digitally inclined consumers encountered usability challenges. Speed and usefulness exhibited the same tendency; although they did not attain statistical significance in regression analysis, they were pivotal in decision tree and clustering models. This indicates that although these criteria may not independently propel adoption, they considerably affect customer segmentation and engagement levels.

Security emerged as a crucial factor, with regression analysis affirming its significance, decision trees prioritizing it, and clustering highlighting its specific relevance for suspicious customers apprehensive about digital banking due to trust issues. Conversely, social pressure demonstrated a more intricate and variable link. Hypothesis testing revealed no significant difference between users and non-users, decision trees indicated minimal importance, and regression analysis demonstrated a negative association, meaning that external factors may hinder rather than promote adoption. Clustering redefined social pressure as a trait of reluctant users rather than an incentive for participation. This suggests that individuals embrace online banking based on individual assessments of convenience, security, and efficiency rather than societal pressures.

Income, although not a substantial predictor in regression or decision trees, contributed to segmentation by distinguishing user groups rather than directly affecting adoption. This indicates that financial competence is not a principal obstacle to online banking, yet it may influence the manner in which certain segments engage with digital financial services.

These findings indicate that enhancing usability, security, and transaction speed will be more effective in promoting digital banking adoption than depending on social influence or financial incentives. The segmentation research underscores the necessity for customized interventions: mitigating trust issues for doubtful users, enhancing usability for reluctant traditionalists, and utilizing tech-savvy adopters as proponents of digital banking.

5.1. Theoretical and Practical Implications

This study elucidates the elements affecting online banking adoption, emphasizing the interaction among PEU, SEC, SPD, SP, and consumer segmentation. The Technology Acceptance Model (TAM) paradigm received robust acceptance, with simplicity of use identified as the paramount factor influencing online banking uptake. All analytical methods—logistic regression, decision trees, and clustering—indicate that consumers who saw internet banking as intuitive and user-friendly exhibited a markedly greater propensity for adoption. Both regression and decision tree analyses converged on perceived ease of use as the most decisive factor in adoption. This dual confirmation underscores that usability is not only statistically significant across the entire sample but also the first conditional criterion separating adopters from non-adopters. This corroborates previous studies suggesting that usability is a crucial determinant in the uptake of digital financial technologies.

Perceived usefulness did not emerge as a statistically significant predictor in regression, yet the decision tree ranked it as a top conditional driver. This suggests that usefulness may not uniformly predict adoption across the population but becomes influential when combined with other factors such as ease of use. In other words, usefulness acts as a threshold variable rather than a universal determinant.

Perceptions of security significantly influenced hesitant users. Regression research identified security as a significant predictor, decision trees positioned it among the foremost determinants, and clustering revealed that suspicious and hesitant consumers identify trust issues as an impediment to adoption. This underscores the significance of fraud prevention protocols, clear communication, and cybersecurity improvements in digital banking strategy.

From a practical perspective, financial institutions can translate these findings into several strategies. Given that ease of use has become the predominant factor, banks must concentrate on enhancing user interfaces, optimizing navigation, and providing intuitive digital experiences. Moreover, they can implement simplified mobile interfaces, tutorials for new users uploaded on their website or social media accounts and multilingual support through customer support services to reach less digitally skilled populations. Considering the substantial impact of security apprehensions among reluctant consumers, strategies must prioritize fraud prevention measures, clear communication, and strong cybersecurity protocols to enhance customer trust. In order to strengthen trust, banks should invest in transparent communication regarding security matters and include multiple fraud detection tools.

Despite speed and usefulness lacking statistical significance in regression analysis, their relevance in decision tree and clustering analyses indicates that banks should prioritize real-time transactions, seamless integrations with e-commerce platform.

Interestingly, social pressure had a complex and contradictory role. Hypothesis testing and decision tree models suggested that external influences do not significantly drive adoption, and regression analysis even indicated a negative association, meaning that perceived social coercion may discourage use rather than encourage it. Clustering further reinforced that hesitant users experience social pressure, but it does not translate into engagement. This suggests that financial institutions should avoid persistent campaigns and instead focus on personal benefits, autonomy, and convenience rather than peer-driven marketing strategies.

The consumer segmentation analysis revealed four distinct groups with varying adoption behaviors:

• Tech-Savvy, Highly Engaged Users (48%)—Confident users who prioritize usability, speed, and security. This group represents the ideal audience for advanced digital features, AI-driven services, and personalized financial tools.
• Moderate Digital Banking Users (27%)—Users with mild security concerns who require reassurances about fraud protection and improved customer support.
• Skeptical Digital Users (14%)—Hesitant users who struggle with usability and security concerns. This group would benefit from educational initiatives and simplified digital experiences.
• Reluctant Traditionalists (11%)—Strongly resistant to digital banking, viewing it as overly complex and risky. This segment may require gradual onboarding, hybrid banking options, and targeted financial incentives to encourage adoption.

Since income was not a major determinant of adoption, expanding accessibility through financial incentives is unlikely to drive significant behavioral change. Instead, efforts should be concentrated on usability improvements and trust-building measures that resonate across different income levels. These insights suggest that banks should move beyond broad marketing efforts and instead implement user-centric, trust-focused, and efficiency-driven strategies to make digital banking indispensable in everyday financial management.

The Extended TAM centers on direct effects from perceptions (PEU, PU, SEC, SPD, SP, AI, ADAPT) and demographics (INC, EDU, AGE) to Intention (BI) and Adoption (IB_USE), and from Adoption to Online Purchasing Decisions (OPD). In logistic regression, PEU and SEC were the strongest direct predictors of IB_USE, while SP had a significant negative effect; INC was marginal (p ≈ 0.055). The CART highlighted SPD and PU importance via non-linear thresholds, complementing the regression. Hypotheses H1–H2, H4, H8, H10–H12 were supported in bivariate tests; H9 was not supported (and reversed in multivariate). H3.1–H3.2 were supported, linking adoption (IB_USE/MB_USE) to OPD. This resolves earlier inconsistencies by aligning the model, hypotheses, measures, and the two predictive approaches.

5.2. Future Research Directions

Future research should explore longitudinal trends in online banking adoption, tracking how consumer behavior evolves with technological advancements and regulatory changes. Investigating the impact of AI-driven banking, FinTech innovations, and cybersecurity measures can provide insights into consumer trust and adoption patterns. Additionally, cross-cultural comparisons could reveal how factors such as digital infrastructure, economic conditions, and regulatory policies shape banking preferences across different regions. Given the rise of cryptocurrencies and decentralized finance (DeFi), future studies should also examine how these emerging financial technologies interact with traditional banking services.

Further research should focus on behavioral and psychological drivers of adoption, such as risk perception, digital literacy, and emotional responses to technology. Studying the effectiveness of personalized banking experiences, gamification, and incentive-based engagement could provide strategies for improving user retention. Moreover, addressing financial inclusion challenges by studying underrepresented groups—such as low-income populations, the elderly, and individuals with limited digital access—can help bridge the digital divide and ensure equitable access to financial services in an increasingly digital economy.

Future studies can enhance digital banking strategies, improve consumer trust, and foster financial inclusion in an increasingly digitalized economy.

6. Conclusions and Limitations

Digitalization in banking is more than just a technological shift—it is a fundamental change in how financial services interact with consumers and support economic growth. By fostering digital skills and integrating advanced banking technologies, financial institutions not only enhance customer experiences but also drive the expansion of e-commerce, creating a more connected and financially inclusive world.

The findings offer several theoretical contributions. First, they extend the Technology Acceptance Model (TAM) by validating usability and security as the dominant predictors of adoption, while highlighting the limited and sometimes counterproductive role of social influence. While speed and usefulness influence engagement, they are secondary to consumer trust and ease of navigation. The role of social pressure is minimal or even counterproductive, indicating that banks should focus on individual benefits rather than external influence campaigns.

Second, the use of complementary analytical methods—logistic regression, decision trees, and cluster analysis—demonstrates the value of combining linear and non-linear approaches in order to capture both direct effects and complex interaction patterns. Third, the segmentation analysis revealed four distinct consumer profiles, underscoring the importance of integrating behavioral and demographic dimensions into adoption models. Together, these contributions advance existing theory by refining the Extended TAM framework for digital banking adoption in emerging economies.

Taken together, these results show that logistic regression isolates the strongest linear effects (i.e., ease of use and security) while the decision tree uncovers conditional pathways, highlighting the importance of usefulness and speed. Far from being contradictory, the two methods complement one another: regression demonstrates the universal drivers of adoption, whereas decision trees reveal context-dependent factors that segment users into distinct adoption profiles. The study contributes to the existing literature by combining classical statistical techniques with machine learning-based decision trees and cluster analysis, offering a multi-dimensional perspective on user attitudes toward digital banking. While the research focused on a specific demographic segment, it establishes the foundation for broader, cross-generational studies and cross-national comparisons in future work.

From a practical standpoint, the findings can be used as a guiding framework for banks and financial institutions seeking to improve or expand their digital services in Romania or in comparable emerging markets. Strengthening platform usability, ensuring robust data security, and reducing social pressure-based messaging can lead to higher engagement and customer satisfaction.

Moreover, the segmentation analysis revealed distinct consumer profiles, each with varying expectations, levels of trust, and degrees of digital engagement. These insights emphasize the importance of personalization and targeted communication strategies in digital service design and marketing.

Despite its contributions, this study has several limitations. First, the sample may not be fully representative of all consumer demographics, particularly those with limited access to digital banking infrastructure. Future research should incorporate a more diverse sample across different socioeconomic backgrounds and geographic regions.

The survey was conducted exclusively in Romania and targeted individuals between 19 and 39 years old—a demographic group identified as digitally literate and financially active. Therefore, future studies should employ improved methodologies for participant selection, such as stratified or random sampling, and examine a broader range of age demographics or diverse countries. Furthermore, integrating objective behavioral data (e.g., digital usage metrics) with self-reported perceptions will enhance the validity of the findings.

Second, the study relied on self-reported perceptions and behaviors, which may be subject to response bias. Future research could incorporate behavioral tracking or experimental designs to measure actual online banking usage patterns more objectively.

Third, while the analysis identified key predictors of adoption, it did not explore longitudinal changes in consumer behavior. A future study could track adoption trends over time, identifying how attitudes and engagement evolve with new digital banking innovations. Longitudinal research may elucidate the evolution of consumer attitudes in response to technology advancements and regulatory changes, especially as AI, blockchain, and open banking transform the financial landscape.

Finally, emerging factors such as AI-driven banking services, blockchain security, and personalized financial management tools were not extensively analyzed. Future research should examine how these innovations influence digital banking adoption in different consumer segments.

Despite these limitations, the findings offer valuable insights for financial institutions seeking to optimize their digital banking strategies. By focusing on usability, security, and consumer-centric engagement, banks can drive adoption while enhancing the overall customer experience in an increasingly digital financial landscape.

This research highlights that enhancing usability and security, while reducing reliance on social influence, is critical to promoting online banking adoption. By combining advanced analytical methods with segmentation insights, the study offers both theoretical refinement and practical guidance for banks seeking to navigate the challenges of digital transformation. These findings not only inform digital banking strategies in Romania but also provide a foundation for future comparative and longitudinal research that will further illuminate the dynamics of consumer adoption in the rapidly evolving digital economy.