1. Introduction
Advancements in digital technologies have reshaped businesses (Reference Klein and TodescoKlein & Todesco, 2021). In certain industries (e.g., music) and among major digital companies (Netflix), digital technologies have completely transformed market dynamics (Reference Soto-AcostaSoto-Acosta, 2020). Traditional companies, in response, have tried to accelerate their digitalization paces, adapting their business models to embrace an ambidexterity strategy that combines physical and digital capabilities (Reference Soto-AcostaSoto-Acosta, 2020).
A notable phenomenon associated with digitalization is the proliferation of startups, which are critical in driving economic development, particularly in industries such as technology, health, and finance (Startup Genome, 2020). Many startups leverage digital technologies to implement novel business models that enhance resource use and efficiency. In these cases, digital technologies trigger fundamental changes in business and generate revenues (Reference Verhoef, Broekhuizen, Bart, Bhattacharya, Qi Dong, Fabian and HaenleinVerhoef et al., 2019).
Digital technologies have also influenced the relationship between companies and their customers (Reference Crittenden, Crittenden and CrittendenCrittenden et al., 2019; Reference Klein and TodescoKlein & Todesco, 2021). These technologies enable companies to reach and engage many customers through social media and digital communication channels (Reference Crittenden, Crittenden and CrittendenCrittenden et al., 2019), creating opportunities for building closer relationships with them (Reference Ansong and BoatengAnsong & Boateng, 2019; Reference Grover and KohliGrover & Kohli, 2013).
Customer engagement (CE) is crucial for digital businesses (Reference Klein and TodescoKlein & Todesco, 2021). CE is defined as the behavioral manifestations of customers toward a brand or firm that go beyond purchasing, driven by motivational factors (Reference Van Doorn, Lemon, Mittal, Nass, Pick, Pirner and VerhoefVan Doorn et al., 2010). It encompasses ongoing communication between companies and their customers, amplifying customers’ physical, psychological, and emotional investments in the company or brand (Reference Brodie, Hollebeek, Jurić and IlićBrodie et al., 2011). Given its significance, CE is linked to various marketing outcomes, including customer satisfaction (Reference Youssef, Johnston, AbdelHamid, Dakrory and SeddickYoussef et al., 2018), loyalty (Reference BakhtievaBakhtieva, 2020; Reference Crittenden, Crittenden and CrittendenCrittenden et al., 2019; Reference Garg, Gupta, Dzever, Sivarajah and KumarGarg et al., 2020), customer rapport (Reference Chen, Weng and HuangChen et al., 2018), and the acquisition of new customers.
In digital businesses, fostering CE has become a strategic priority and a critical driver of long-term success (Reference Brodie, Hollebeek, Jurić and IlićBrodie et al., 2011). Digital startups should craft and implement strategies to effectively connect with and engage their customers in the virtual realm. For example, leveraging data-driven insights and utilizing diverse communication channels can help create personalized experiences that strengthen customer relationships (Reference Verhoef, Broekhuizen, Bart, Bhattacharya, Qi Dong, Fabian and HaenleinVerhoef et al., 2019).
While the entrepreneurship literature has explored various aspects of digital business dynamics, such as digital entrepreneurship development (Reference Jafari-Sadeghi, Vahid, Garcia-Perez, Candelo and CouturierJafari-Sadeghi et al., 2021; Reference LadeiraLadeira et al., 2019), it has yet to examine the role of CE in digital businesses comprehensively. Specifically, there is a lack of a consolidated understanding of how digital startups perceive and address the impact of CE. From a design perspective, this gap aligns with the importance of creating customer-centric digital solutions that enhance engagement and business performance.
This study aims to empirically assess the relationship between customer-related digitalization factors (DIF), CE, and digital business performance (DBP). A structural model was developed to depict the interactions between DIF, CE, and DBP. Furthermore, the mediating role of CE was analyzed. Data was collected through a cross-sectional survey of 125 Brazilian digital startups and evaluated using Partial Least Squares Structural Equation Modeling (PLS-SEM). Brazilian startups are notable for their emphasis on digitalization and customer-centric strategies, which they leverage to achieve market differentiation (Reference da Rosa, Schreiber, Schmidt and Kuhn Juniorda Rosa et al., 2017).
This study’s contributions are twofold. First, it advances existing research by emphasizing a customer-centric perspective in digitalizing business models, shedding light on how digital startups can leverage customer engagement to enhance their business strategies. Second, it highlights the mediating role of CE in the relationship between digitalization and digital business performance, offering new insights into how CE influences the outcomes of digital transformation.
This paper is structured as follows: Section 2 outlines the research hypotheses and presents the conceptual model. Section 3 describes the research methodology, followed by the presentation of results in Section 4. Section 5 discusses the findings, and Section 6 offers a conclusion of this study.
2. Hypothesis development
2.1. Digitalization and business performance
The literature presents different customer-related digitalization factors that influence CE and shape the dynamics of digital businesses within the entrepreneurship context. Four key factors that impact the relationship between CE and digital business are outlined in Table 1. Essentially, companies should tailor their strategies and practices based on these factors to foster long-term relationships with customers.
Customer-related digitalization factors
DIF has significantly influenced how companies retain, co-create, and communicate with their customers. However, anecdotal evidence suggests that many businesses still encounter what is known as the “digitalization paradox” (Reference Gebauer, Fleisch, Lamprecht and WortmannGebauer et al., 2020; Reference GuoGuo et al., 2023). This paradox refers to a situation where companies invest heavily in digitalization and digital offerings but struggle to achieve the anticipated revenue growth despite the well-documented potential of digital technologies to drive growth (Reference GuoGuo et al., 2023). Several studies (e.g., Reference Gebauer, Fleisch, Lamprecht and WortmannGebauer et al., 2020; Reference GuoGuo et al., 2023) have highlighted this issue, attributing it to challenges in managing business models and a lack of customer-centricity as primary contributing factors. Therefore, to explore the impact of customer-related digitalization factors on digital business performance, we propose the following hypothesis:
-
H1 - Customer-related Digitalization Factors (DIF) positively impact Digital Business Performance (DBP).
2.2. Digitalization and customer engagement
While improving CE through digital technologies is widely regarded as a key component of a customer-centric strategy, the impact of DIF on CE remains underexplored (Sashi, 2020). To investigate the influence of DIF on CE, we propose the following hypothesis:
-
H2 - Customer-related Digitalization Factors (DIF) positively impact Customer Engagement (ENG) in Digital Businesses.
2.3. Customer engagement and digital business performance
Literature suggests CE significantly influences financial and non-financial business performance (Reference Chen, Weng and HuangChen et al., 2018; Reference Santini, Ladeira, Pinto, Herter, Sampaio and BabinSantini et al., 2020). CE is even more critical for maintaining competitiveness in digital businesses, as it directly impacts firm performance, behavioral intentions, and word-of-mouth (Reference Santini, Ladeira, Pinto, Herter, Sampaio and BabinSantini et al., 2020). Additionally, CE reflects the co-creation process between providers and customers, which has financial implications for the organization and its customers (Reference Brodie, Hollebeek, Jurić and IlićBrodie et al., 2011). While the performance outcomes of CE have been explored in the literature, this study focuses on the potential mediating role of CE between DIF and DBP. As highlighted earlier, it is crucial to clarify the impact of DIF on DBP (Reference Gebauer, Fleisch, Lamprecht and WortmannGebauer et al., 2020; Reference GuoGuo et al., 2023). This investigation extends the existing research by considering CE as a mediator in this relationship. To provide empirical evidence on the interplay between DIF, CE, and DBP, we propose the following hypotheses:
-
H3 - Customer Engagement (ENG) positively impacts Digital Business Performance (DBP).
-
H4 - Customer Engagement (ENG) mediates the relationship between Digitalization Factors (DIF) and Digital Business Performance (DBP).
3. Research methodology
3.1. Data collection and sample
We conducted a survey following the procedures Reference ForzaForza (2002) and Reference Malhotra and GroverMalhotra and Grover (1998) suggested. Brazilian digital startups were selected for this study due to their significant role in innovation and growth within the Brazilian market (Reference CarriloCarrilo, 2020). We analyzed reports of active startups and innovation ecosystems in Brazil to identify potential participants, compiling contact information for 968 startups. We then sent invitation emails to these startups, utilizing the Qualtrics platform to distribute the survey link over five months, with reminder emails to increase response rates.
After excluding incomplete or inconsistent responses, 125 valid responses were retained, meeting the sample size requirements of three methods: (1) a sample size ten times greater than the number of internal correlations (Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016), (2) the minimum R-squared method (Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016; Reference Kock and HadayaKock & Hadaya, 2018), and (3) the inverse square root method (Reference Kock and HadayaKock & Hadaya, 2018), confirming the adequacy of the sample for analysis.
The survey questionnaire was divided into two sections. The first section included multiple-choice questions designed to characterize digital startups. The startups in our sample spanned 15 sectors, including healthcare (26.40%), information technology (IT) (18.40%), and energy (9.60%). Most respondents held senior management (65.60%) or director (24%) positions, making them key informants. And 60% of the startups operated in the Business-to-Business (B2B) market.
The second section of the questionnaire focused on the proposed conceptual model. Respondents rated their agreement using an eleven-point Likert scale, ranging from 0 (“strongly disagree”) to 10 (“strongly agree”), as recommended by Reference Wu and LeungWu & Leung (2017).
To ensure the quality of the measure, we followed the guidelines of Reference Sarstedt, Hair, Cheah, Becker and RingleSarstedt et al. (2019), using an extensive theoretical background and seeking expert feedback. Specifically, two rounds of validation were conducted. Seven specialists - four from academia and three from industry - reviewed the questionnaire for (i) the content of the theory-related questions and (ii) the clarity of wording and item purpose to ensure respondents could easily comprehend each question. Based on their feedback, we made improvements and conducted a pilot test with five digital startups to validate the questionnaire further.
3.2. Operationalization of variables
Construct scales were developed following Hinkin’s (1988) recommendations. Multiple measurement items were considered for each research construct and selected based on the relevant literature. The independent construct, DIF, is a higher-order construct comprising four first-order constructs: data security (DS), transparency (TR), customer review (CR), and effective communication (EC), as outlined in Section 2.1. The dependent construct, DBP, is a first-order reflective construct that comprises six items based on different performance measurements. Similarly, the mediating variable, “engagement” (ENG), is also a first-order construct, referring to loyalty (or retention), communication, and co-creation. Table 2 provides the manifested variables associated with each construct.
Constructs and variables
3.3. Data analysis
We used PLS-SEM for quantitative data analysis due to its suitability for handling non-parametric statistics and small samples, providing accurate insights into the relationships among the constructs (Reference Hair, Ringle and SarstedtHair Jr et al., 2011, Reference Hair, Risher, Sarstedt and Ringle2019). PLS-SEM is widely accepted in various research areas, including marketing and business (Reference Hair, Sarstedt, Hopkins and KuppelwieserHair Jr et al., 2014).
Data analysis followed the two steps outlined by Reference Hair, Risher, Sarstedt and RingleHair et al. (2019) and Reference Hair, Hult, Ringle and SarstedtHair Jr et al. (2016), supported by SmartPLS© software (Reference Ringle, Wende and BeckerRingle et al., 2015). We assessed the measurement model for reliability and validity in the first step. In the second step, we evaluated the structural model using parameters such as collinearity (VIF), determination coefficient (R2), effect size (f2), and predictive relevance (Q2) (Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016). Hypotheses were validated or refuted based on path coefficients and significance levels (Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016).
3.4. Reliability and validity
We assessed the measurement model for reliability and validity. For reliability, we evaluated item consistency using Composite Reliability, with values ranging from 0.819 to 0.899, surpassing the recommended threshold of 0.70 (Reference Hair, Ringle and SarstedtHair et al., 2011, Reference Hair, Risher, Sarstedt and Ringle2019; Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016). For convergent validity, we examined outer loadings, and the Average Variance Extracted (AVE) (Reference Fornell and LarckerFornell & Larcker, 1981). Most outer loadings exceeded the 0.70 threshold, and those between 0.40 and 0.70 were acceptable as they did not affect the final AVE value (Reference Hair, Risher, Sarstedt and RingleHair et al., 2019; Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016). Additionally, AVE values exceeded 0.50, indicating good convergent validity (Table 3).
Reliability, convergent validity, and divergent validity
Note 1: The diagonal contains the AVE square root, higher than the correlation among variables.
Discriminant validity refers to the extent to which a construct is distinct from other constructs (Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016). We assessed this using two criteria: Fornell and Larcker’s criterion (1981), which compares the square root of AVE with the correlations of latent variables, and the HTMT (Heterotrait-monotrait ratio) criterion, which should be below 0.85 (Reference Hair, Risher, Sarstedt and RingleHair et al., 2019). In our study, the square root of AVE values exceeded the correlations (Table 3), and all HTMT values were below 0.85 (Table 4), confirming the discriminant validity of our model.
HTMT’s criterion analysis
4. Results
We evaluated the structural model using the two-stage disjoint procedure for second-order constructs, as Reference Sarstedt, Hair, Cheah, Becker and RingleSarstedt et al. (2019) recommended. The results are presented in Figure 1 and Table 5. The first step involved checking for collinearity by analyzing the inner variance inflation factor (VIF) values for the predictor constructs that form the DIF construct (Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016). As shown in Table 5, all VIF values are below the conservative threshold of 3.0 (Reference Hair, Risher, Sarstedt and RingleHair et al., 2019), indicating that our model is free from collinearity issues.
We assessed the model’s predictive capacity and the constructs’ relationships (Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016). According to Cohen (1988), the determination coefficient (R2) values indicate the predictive strength of the model: substantial (R2 > 0.26), moderate (0.13 < R2 < 0.25), and weak (R2 < 0.12) for social sciences. Table 5 shows a moderate influence of DIF on the exogenous construct DBP (R2 = 23.9%, R2 Adj = 22.6%). However, our model has limitations in predicting the behavior of the ENG construct (R2 = 0.047), suggesting that other unexamined variables may influence the variance of ENG.
In addition to R2 values, we also examined the effect size (f2) and predictive relevance (Q2). The effect size values in our model range from 0.049 to 0.166, with values above 0.02 considered medium and those above 0.15 considered high (Reference Hair, Hult, Ringle and SarstedtHair Jr et al., 2016). For Q2, values greater than zero indicate predictive relevance (Reference Hair, Ringle and SarstedtHair et al., 2011, Reference Hair, Risher, Sarstedt and Ringle2019). We obtained Q2 values of 0.107 for DBP and 0.029 for ENG, with an omission distance of seven, confirming that our model demonstrates predictive relevance.
The path coefficients and corresponding significance levels (p-values) were calculated using a bootstrapping procedure with 5,000 subsamples (Reference Hair, Hult, Ringle and SarstedtHair et al., 2016). The results, presented in Table 5, confirm the validation of the following hypotheses: H1: DIF → DBP (β = 0.364, p-value = 0.000); H2: DIF → ENG (β = 0.216, p-value = 0.013); and H3: ENG → DBP (β = 0.256, p-value = 0.003). Additionally, the indirect mediating effect of customer engagement (ENG) on the relationship between DIF and DBP was assessed and confirmed (β = 0.055, p-value = 0.100). Both the direct and indirect effects of DIF on DBP are significantly positive, suggesting mediation of CE in the relationship between DIF and DBP.
Structural model
Hypotheses Analysis
Note 1: mediation effect is supported by a significance level of 10%.
5. Discussion
Our study confirms that Data Security, Transparency, Customer Review, and Effective Communication are interrelated and form the higher-order construct of DIF. This finding extends previous research that has examined individual digitalization factors and their impact on DBP, offering an integrative perspective that provides deeper insights into the mechanisms by which DIF influences DBP. Building on previous studies emphasizing the technological and organizational aspects of digitalization and digital transformation (e.g., Reference Verhoef, Broekhuizen, Bart, Bhattacharya, Qi Dong, Fabian and HaenleinVerhoef et al., 2019), our study contributes to the existing literature by focusing on customer-related digitalization factors.
Our results indicate that DIF directly and positively impacts DBP, suggesting that these factors should be carefully considered in digital business models and effectively implemented in business processes associated with technological infrastructure to ensure successful digital business outcomes. Furthermore, DIF positively affects CE, as confirmed by hypothesis H2 (Table 5). When digitalization factors are implemented with a customer-centric approach, they enhance customer engagement. For instance, Data Security and Transparency, while primarily legal requirements are perceived by customers as signals of trust and concern for their data protection. Similarly, Customer Review and Effective Communication, common in digital businesses, can significantly boost performance by fostering a customer-centric approach.
Another notable finding from our study is the mediating role of CE, with a significance level of 10% (β = 0.055, p-value = 0.100). CE has been primarily associated with loyalty, communication, and co-creation. In the context of digital startups, CE plays a crucial role in bridging digitalization and business performance. Particularly in digital startups, CE is even more vital and nuanced than in traditional businesses. While literature supports the positive impact of CE on performance (Reference Chen, Weng and HuangChen et al., 2018; Reference Garg, Gupta, Dzever, Sivarajah and KumarGarg et al., 2020), our study uniquely validates its mediating role in the relationship between digitalization and business performance. CE manifests in customers’ emotional connection and proactive involvement with digital businesses. When digitalization drives CE, customers are more likely to contribute positively to the company, enhancing performance.
The higher the implementation level of digitalization factors (Data Security, Transparency, Customer Review, and Effective Communication), the better the business performance, particularly in sales and customer acquisition. The study proposes this virtuous cycle: as digitalization factors increase, customer satisfaction rises, leading to greater engagement, more suggestions, and new customers, improving performance. Thus, leveraging digitalization factors from the CE perspective boosts business performance by enhancing the customer experience and fostering new customer acquisition.
6. Conclusion
This study empirically assesses the positive relationship between DIF, CE, and DBP. The relationship was evaluated using a conceptual model and a cross-sectional survey of 125 Brazilian digital startups, analyzed through Partial Least Squares Structural Equation Modeling (PLS-SEM).
The study offers two main theoretical contributions. First, it advances the understanding of the interplay between digitalization and CE by demonstrating that digitalization factors positively impact business performance when viewed from a customer-centric perspective. Second, the study provides a theoretical model identifying positive and significant relationships between DIF, CE, and DBP. This conceptual model is a foundation for future research in digital business and CE.
From a practical perspective, this study offers valuable insights for digital startups. It positions customer engagement as a key driver of market and financial success, which can elevate DBP, aligning with the virtuous circle identified in this study. Based on our findings, digital startups should prioritize customer relationships, particularly through social media engagement.
Furthermore, this study contributes to the field of design by highlighting the importance of customer-centric digital strategies. The findings underscore the role of design in shaping digital interactions, optimizing user experiences, and structuring business models that foster engagement and performance. By integrating digitalization and CE, this study provides a framework for designing more effective, scalable, and sustainable digital business strategies.
While this study makes important contributions, it also presents limitations that could be explored in future research. First, our sample is limited to Brazilian startups, so the findings may not be generalizable to all types of businesses, especially traditional companies. Future studies could test the conceptual model in other sectors or businesses undergoing digital transformation. Second, the validated conceptual model serves as an initial proposal, and further research could refine and expand upon the variables not covered in this study. Third, while our study uses quantitative data, qualitative approaches could offer deeper insights into the relationship between digitalization and customer engagement in digital business contexts. Lastly, the study is based on Brazilian startups, and future research could compare startups from different countries to identify cross-cultural differences.
Acknowledgments
The authors would like to sincerely thank the Higher Education Personnel Improvement Coordination (CAPES) and the National Council for Scientific and Technological Development (CNPq), from Brazil, for their financial support of this research.
Open access
