Precursors of interactional value formation through digital banking interactions

Relationship quality

In dyadic interactions involving two or more parties, relationships or partnerships form the foundational base for value co-creation (Ballantyne and Varey, 2006). This relationship dynamic is particularly significant in banking interactions, which involve sensitive and confidential information where trust, commitment, and connection are paramount (Neghina et al., 2013). Ballantyne and Varey (2006) highlight the integral role of relationships, communication, and knowledge in Interactive Value Formation (IVF), while Mukherjee and Nath (2003) support this framework using trust-commitment theory. Trust, as a relationship variable, has been shown to co-create value across industries, as evidenced by Garanti (2023) in tourism interactions and Järvi et al. (2018) in qualitative studies exploring trust in co-creation processes. In banking, strong relationship quality reduces customer anxiety and fosters stability in interactions (Morgan and Hunt, 1994). This is supported by Cambra-Fierro et al. (2019), who found that customers with relationship-oriented behavior enhance co-creation in online banking. Similarly, Osei-Frimpong et al. (2017) used the critical incident technique to validate the role of trust in enabling a safe environment for patient-doctor information sharing, which can be analogized to the banking sector’s trust-intensive interactions. Emotional connections between employees and customers also play a pivotal role in co-creation (Salovey and Mayer, 1990), as observed in tourism and healthcare contexts (Teng and Tsai, 2020; Tari Kasnakoglu, 2016).

Despite the increasing digitalization of banking services, physical interactions remain crucial, particularly for complex financial solutions, where customers tend to trust human advisors over automated or robotic alternatives (Gremler and Gwinner, 2008). However, contradictory findings suggest that the relationship between trust and IVF is not universally positive. For instance, Gyllenhammar et al. (2023) observed an inverse relationship in public insurance services, where excessive trust may complicate processes like claim disbursement. Similarly, Mostafa and Ibrahim (2020), in an Egyptian banking context, reported no significant role of trust in customer co-creation.

These mixed findings underscore the complexity of relationship dynamics in IVF, suggesting that while trust and relationship quality are generally positive enablers of co-creation, their impact may vary based on contextual and cultural factors. Further research is needed to explore these nuances, particularly in digital banking, where trust mechanisms are evolving with technological advancements.

Therefore, we hypothesize that H1: Relationship quality positively influences IVF.

Communication

Ballantyne and Varey (2006), Gustafsson et al. (2012), and Neghina et al. (2013) emphasize the critical role of communication in co-creation, proposing that bidirectional and transparent interaction fosters equal and meaningful participation from all parties. Empirical evidence from various industries substantiates the positive relationship between communication and Interactive Value Formation (IVF) (Smith, 2013; Kashif and Zarkada, 2015; Malar et al., 2019; Yin et al., 2019; Echeverri and Salomonson, 2017; Chathoth et al., 2020; Keeling et al., 2020; Sthapit and Bjørk, 2020; Guan et al., 2022).

In digital banking, the absence of face-to-face interaction amplifies the importance of clear, bidirectional communication channels. The principles outlined by Ballantyne and Varey (2006), Gustafsson et al. (2012), and Neghina et al. (2013) are particularly relevant, as communication in this context must compensate for the lack of physical cues and traditional customer service interactions. Bidirectional communication ensures that both customers and banks engage in meaningful dialogue, fostering trust and co-creating value. In banking, communication gaps between employees and customers can significantly hinder value creation (Kashif and Zarkada, 2015). Online banking requires dynamic and effective communication to facilitate co-creation (Malar et al., 2019). Evidence from healthcare interactions underscores this need, as Keeling et al. (2020) highlight the positive impact of bidirectional communication on co-creation outcomes. Similarly, studies in the hospitality sector reveal the importance of communication, with Chathoth et al. (2020) employing quantitative methods and Boadi et al. (2020) and Sthapit and Bjørk (2020) using netnography approaches to confirm its role in enhancing customer co-creation.

Additional insights come from public transport services, where Echeverri and Salomonson (2017) and Guan et al. (2022) demonstrate a strong link between dialogue and IVF. These findings collectively highlight that effective communication is a universal driver of value co-creation across service industries, reinforcing its importance in the banking sector. Prioritizing interactive dialogue in digital banking contexts can bridge communication gaps and enhance the co-creation experience for customers and service providers alike.

Therefore, we hypothesize that H2: Communication positively affects IVF.

Knowledge

Prahalad and Ramaswamy (2004a, b) and Neghina et al. (2013) identify three critical subconstructs of Interactive Value Formation (IVF): information sharing, information seeking (Nonaka and Takeuchi, 1995), and feedback. Ballantyne and Varey (2006) further validated the significance of these dimensions through their conceptual framework. Despite this foundation, the literature predominantly focuses on employees’ perspectives regarding knowledge exchange (e.g., Chathoth et al., 2023; Guan et al., 2022; McNaughton et al., 2002), often underemphasizing the critical role of customer participation. Studies from diverse sectors underscore the positive relationship between customer engagement in knowledge-sharing activities and value co-creation (Aarikka-Stenroos and Jaakkola, 2012; Agrawal and Rahman, 2015). For instance, Chaudhuri et al. (2023) employed quantitative methods to demonstrate the beneficial role of knowledge sharing in complex healthcare services. Similarly, in the post-service stage of the tourism industry, feedback emerges as a pivotal factor (Codá et al., 2024).

Customer information-sharing traits also positively influence co-creation processes across various service domains. Wu et al. (2023) highlight this in alignment with findings by Järvi et al. (2018). Additionally, Guan et al. (2022) investigated the role of information symmetry, presenting evidence for its positive impact on IVF. However, this perspective has faced critique; Osei-Frimpong et al. (2015) observed that in healthcare, an elevated display of knowledge by patients could adversely affect interactions with physicians, leading to knowledge conflicts. Such conflicts underscore the nuanced and context-dependent nature of customer knowledge contributions to IVF. This complexity necessitates further exploration into how knowledge-sharing dynamics influence co-creation processes across different service contexts.

Based on the review, we hypothesize that H3: Knowledge positively affects IVF.

Ethical considerations

Before administering the survey questionnaire, written informed consent was obtained from all participants. The consent process began with an information sheet that outlined the research objectives, expected outcomes, and potential implications, followed by a section for participants to provide their informed consent. The form assured participants that all data collected would be used exclusively for research and publication purposes, with strict confidentiality maintained for personal information (see Extended Data, Ramya, 2025). Additionally, ethical approval for this empirical study in the Indian Banking industry was obtained from the Manipal Academy of Higher Education, India, Ethics Committee vide reference number MAHE/CDS/PHD/2021 dated 29^th Nov’21.

Research design

A cross-sectional quantitative survey-based approach was adopted to collect data were collected over the period from 2022-2023 using a snowball sampling approach. The independent variables taken as “precursors” of IVF are relationship quality such as trust, connection and commitment, communication referring to direction, content and frequency, and knowledge that involves information sharing and feedback; the dependent variable is IVF.

Measurement tools/scales

The questionnaire was prepared by adopting an established scale adapted from Neghina et al. (2013) on precursors such as relationship quality, communication, and knowledge. Yi and Gong’s (2013) scale was adapted for measuring co-creation in digital banking interactions. Different types of Likert scales were used (7-point and 5-point) to prevent response bias and obtain reliable results. The survey questionnaire can be found as Extended data (Ramya, 2025).

To provide a holistic view of the customer’s perspective, we have included the digitally active customers of the private sector (ICICI Bank and Axis Bank), public sector (SBI and Bank of Baroda), and small finance banks (AU Small Finance Bank) of the Indian banking system. We visited the regional offices of the selected banks and requested the management to use us as an opportunity to obtain customer’s perceptions of digital interactions. Post approval, we met the customers in the respective branches who were digitally active in bank interactions. The study considered interactions through mail, phone, chatbots, WhatsApp, and text messages as digital interactions. To ensure anonymity, a pilot survey was conducted to refine the process and identify potential challenges for the final study. The pilot sample comprised 10% of the final dataset of 442 participants and was selected using the snowball sampling technique. The initial list of participants was asked to provide three customer references, who were then contacted via email to obtain consent for participation. Those who agreed were further asked to share three additional references, enabling the collection of the final sample after applying the necessary filtration criteria. To minimize response bias in the final survey, we employed concise questions, reverse-coded items, and avoided leading language in the questionnaire.

Sample size

The sample size consisted of 442 digitally active bank customers of private, public, and small finance banks in the Karnataka region. Survey questionnaires were shared online with the bank customers.

Data analysis

PLS-SEM 4.0 was employed to assess the measurement and structural model (Ringle et al., 2024). The copyright license for the SmartPLS software was bought under order ID 2024-06155 dated 19^th Aug’24. The measurement model establishes the reliability and validity of the constructs, and the structural model determines the significance of the hypothesized relationships (Hair et al., 2013; Henseler et al., 2012). Alternatively, JASP, an open-source software, can be used to conduct Structural Equation Modeling (SEM) through the Iavaan R package, which follows a covariance-based approach.

Internal consistency reliability

A measurement model is said to have acceptable internal consistency reliability when the composite reliability (CR) is higher than 0.7. The CR of each of the indicators ranges between 0.888 and 0.958, which is higher than 0.7. Cronbach’s alpha values are also within the acceptable range, higher than 0.7 (Henson, 2001), for all the constructs. AVE values were higher than 0.5; hence, convergent validity was established (refer to Table 1).

Outer loadings

Outer loadings reflect the absolute contribution of the indicators to the latent variables (Hair, 2013). Based on the outer loadings, RQT1, RQT2, RQT3, RQC1, RQC2, and RQC3 in RQ, C3, C4, C5, C6, C8, C9, C10, and C11 in C, and K2, K3, K4, K5, and K6 in K were retained. RQT1, RQT2, RQT3, RQC1, RQC2, and RQC3 values were higher than 0.7, ranging from 0.777 to 0.963. The indicators of the construct ‘communication’ were also found to contribute to the latent variable, as the outer loadings ranged from 0.799 to 0.929. The ‘knowledge’ indicators were also within an acceptable range, ranging from 0.814 to 0.862. All sub-constructs of IVF except IVFT reported outer loading of less than 0.7, which is 0.669 (Refer to Table 2).

Construct validity

To validate the discriminant validity, tests, namely, average variance extracted (AVE) and HTMT, were observed, and multicollinearity was checked (VIF). To assess the discriminant validity, HTMT values (acceptable range >0.85) were observed. HTMT (heterotrait-monitrait ratio) measures the similarity observed between the latent variables. HTMT values were reported to be lower than 0.90 (Hair et al., 2013); hence, discriminant validity was established between the proposed constructs, RQ, C, and K (Refer to Table 3).

The next test was VIF (variance inflation factor). It assesses the multicollinearity between the indicators. VIF values of 5 or higher indicate possible issues with collinearity problems. The VIF values obtained were below 5, except for RQT2 (9.734), RQT3 (5.922), C5 (5.859) and C10 (6.256), whose VIF values were higher than 5 (refer to Table 4).

IVF was considered a second-order construct, and all four latent variables (IVFA, IVFPI, IVFRB, IVFT) were taken by comprising their respective indicators for the two-stage disjoint approach. It is a reflective-formative construct; hence, outer weights for latent scores were checked. Since all sub-constructs had outer weights lower than 0.5, bootstrapping was carried out and outer loadings were checked. The loadings were found to be above 0.5 and significant. Hence, these sub-constructs were considered in the higher-order construct IVF.

Model fit

Model fit was tested to check the goodness of fit measure of the proposed model. We tested SRMR (<0.08), NFI (>.90), R square (0.75=substantial, 0.50=moderate, 0.25=weak) (Hair et al., 2013), and f square (>=0.02=small, >=0.15=medium, >=0.35=large). The SRMR value was 0.066, and the NFI was 0.872, which validates the goodness of fit. The R² value of 0.549 indicates that the independent variables (RQ, C, and K) moderately impact IVF. Another test conducted was f-square. It explains the impact of latent variables. The values reveal that relationship quality and knowledge have a medium effect on the model, whereas communication has a small effect (Refer to Table 5).

Finally, path analysis was examined. It is used to explain the relationship between the constructs. We conducted bootstrapping to test the statistical significance of the path coefficients. Knowledge (K) was found to have a more significant influence on IVF than RQ, with the path coefficient of the former (0.424) being higher than that of the latter (0.345). In the current study, RQ➔IVF (p=0.000, t=7.073, path coefficient=0.345), Communication➔IVF (p=0.659, t=0.441, path coefficient=0.034) and K➔IVF (p=0.000, t=6.850, path coefficient=0.424) values signify that H1 and H3 are accepted, as RQ and K significantly and positively affect IVF, and H2 is rejected, as communication does not affect IVF.

Limitations and future scope

Although the study incorporates a diverse set of banks, the sample is limited to Indian banks, which may restrict the generalizability of the findings to other geographical contexts. This study utilizes cross-sectional data, which precludes drawing causal inferences between predictor and criterion variables. The study considers the perceptual model that relies on self-reported measures. Another limitation could be that the responses might be influenced by individual biases and social desirability. Future research could expand the scope to validate the proposed framework in different regions, enhancing its applicability across varied banking environments. Furthermore, while this study centers on digital interactions, future investigations could incorporate in-person banking interactions to provide a more comprehensive understanding of Interactive Value Formation (IVF). The present framework examines precursors to IVF, yet there are likely additional factors, such as situational variables and cultural orientations, that could influence this relationship. Future studies are encouraged to explore the moderating and mediating effects of such variables, thereby offering a deeper understanding of the dynamics underlying value formation.