Configurational pathways to professional service innovation: a fuzzy set analysis of provider adoption decisions

The healthcare sector faces mounting structural challenges, including escalating global health expenditures, inefficiencies in service delivery, and increasing pressure to meet growing and complex patient demands. Global health spending reached USD 9.8 trillion in 2022, accounting for 9.9% of the global GDP, underscoring the urgency of improving efficiency and sustainability in healthcare systems (World Health Organization [WHO], 2024). In response to these challenges, service innovation has emerged as a critical mechanism for transforming the design and delivery of healthcare services, with digital transformation playing a central role (Opazo-Basáez et al., 2022; Berry et al., 2020; Berry and Bendapudi, 2007; McColl-Kennedy et al., 2012; Danaher and Gallan, 2016; Figueroa et al., 2019; Fusco et al., 2023). Health information technologies are increasingly reshaping communication processes and care delivery, supporting both clinical effectiveness and public health promotion (Mithas et al., 2020; Urueña et al., 2016). However, despite the recognized potential of digital transformation, the adoption of these innovations depends heavily on healthcare professionals, whose autonomous decision-making, professional norms, and knowledge asymmetries create complex adoption dynamics (Vink et al., 2021). Consequently, the configurational nature of technology adoption among healthcare professionals remains insufficiently understood in the service management literature (Green et al., 2016; Venkatesh et al., 2003; Greenhalgh et al., 2004; Berry and Bendapudi, 2007; Walter and Lopez, 2008; Rahimi et al., 2018; Pilnick and Dingwall, 2011; Barrett et al., 2015; Greco, 2020; Pappas and Woodside, 2021; Cobelli et al., 2021; Garavand et al., 2022; Lambert et al., 2023; Sze et al., 2024).

Technology adoption in healthcare has been widely examined across three interrelated streams of literature: the healthcare system context, users' perspectives, and professionals' perspectives. First, in the healthcare context, particularly in countries with centralized public health systems, digital healthcare adoption is strongly influenced by institutional logics and governance structures. Healthcare is often perceived by professionals as a function of state resources and economic policy rather than service innovation, which constrains their openness to digital solutions (Tafuro and Dammacco, 2022, pp. 91–107). This institutional framing has contributed to resistance toward service innovations, despite their potential to improve efficiency and access for underserved populations (Patrício et al., 2020; Talwar et al., 2023). Second, from the users' perspective, the literature shows that conventional, hierarchical, face-to-face care models remain dominant and are frequently perceived as more empowering and trustworthy than digital alternatives (Alshammari et al., 2022; Zoghlami and Ben Rached, 2024). These perceptions shape demand-side acceptance and indirectly influence organizational adoption. Third, and most critically for this study, technology adoption from the professionals' perspective remains underexplored. Healthcare professionals are central actors in value co-creation (Capponi and Corrocher, 2022), and their attitudes and decisions determine whether decentralized, home-care-oriented digital service models are realized (Hogreve and Beierlein, 2023). Although some studies have examined professionals' propensity to adopt service innovations (de Grood et al., 2016), much of the existing research remains broad in scope (Okpechi et al., 2022; Stanimirovic, 2024), leading to the persistent conclusion that digital service innovation in healthcare is insufficiently implemented (WHO, 2021). To address this gap, our study explores how healthcare professionals perceive digital services by examining the cognitive and emotional mechanisms underlying their motivation to adopt and recommend them to patients. Using focus groups and fsQCA, we identify key facilitators of adoption and highlight the crucial role of professionals who act as facilitators by building competence and confidence while aligning organizational support, incentives, and workflow integration (O'Donnell et al., 2007). As Borges do Nascimento et al. (2023) emphasize, “high-quality evidence supports that training and educational programs, multisector incentives, and the perception of technology effectiveness facilitate the adoption of digital technologies” (p. 1).

This study examines factors influencing healthcare professionals' willingness to adopt service innovations. A working system was presented to 315 professionals, allowing them to interact with the technology and engineers. Quantitative and qualitative data were collected and analyzed using fuzzy-set qualitative comparative analysis (fsQCA), which identifies equifinal combinations of conditions that shape technology adoption decisions (Pappas and Woodside, 2021). This approach reveals how multiple factors combine to enable or inhibit adoption in professional settings.

This study advances service management by shifting adoption from patient-centric to a dyadic, co-creative perspective that foregrounds the provider side of digital healthcare services. Using a configurational approach with self-employed ENTs, we show that professionals' intentions emerge from multiple pathways where professional identity, perceived patient value, organizational readiness, and technology affordances shape decisions. We find that Hedonic Motivation, Performance Expectancy, Price Value, and Social Influence are core ingredients; high adoption intention occurs when these combine with supportive effort expectations and Facilitating Conditions, alongside low Anxiety, covering many cases (Scarpi et al., 2021).

We re-center the provider by making their value calculus visible, showing how clinical efficacy, income continuity, and expectations interplay, extending service-innovation scholarship to co-production in professionalized settings. We introduce configurational logic to explain heterogeneous outcomes. By demonstrating conjunctural causation and asymmetry, this study explains why similar professionals diverge in intentions. Low Anxiety acts as an enabling complement to ease of use and support, refining classical adoption models and reflecting professional autonomy (Engström et al., 2022). We extend adoption theory by showing that for self-employed ENTs, Hedonic Motivation and Price Value are core conditions, while Social Influence operates through peer norms and patient expectations. This broadens UTAUT-family insights by incorporating governance, economic, and identity logics that govern professional service work.

Our findings integrate Expectancy Theory, Self-Determination Theory, and Social Cognitive Theory within a professional service innovation framework. We demonstrate that adoption intentions emerge from configuring expectancy-value calculations (performance gains and price trade-offs), intrinsic motivation (hedonic satisfaction tied to professional identity), normative pressures (Social Influence through peer networks), and emotional regulation (Anxiety management). Unlike consumer technology adoption where single motivations suffice, professional service innovation requires these mechanisms to operate simultaneously, reflecting the high-stakes nature of professional practice decisions. This configurational psychological mechanism advances service innovation theory by showing that provider adoption differs from patient adoption in healthcare service transformation.

We offer guidance for service design and policy by translating “recipes” into key levers: economic viability through pricing that works for professionals; operational integration via training and workflow support that reduce effort; and social proof through peer examples and communication that reinforce norms. These levers enable effective deployment of innovations for home-based care, operational hubs, service integration, accessibility, and remote care models.

These findings advance service-dominant logic by showing how provider value propositions emerge through configurational mechanisms, extending understanding of value creation in professional service ecosystems (Vogus et al., 2021). The configurational approach addresses a gap in service research, where traditional models inadequately capture the pathways of professional service innovation. These contributions reposition service innovation adoption as configurational and managerially designable, moving from technology rollout to service-ecosystem orchestration in digital care (Vink et al., 2021).

Research (e.g. Jacob et al., 2022; Bertolazzi et al., 2024; Tan et al., 2025; Fareed and Kirkil, 2025) has examined technology acceptance and adoption through models such as UTAUT and its extensions. Building on this foundation, this study offers contributions that extend beyond existing frameworks and provides an understanding of technology adoption in healthcare from a service management and innovation perspective.

First, this study focuses on professionals rather than end-users, examining adoption from the service providers' perspective rather than that of patients or consumers. This aligns with service management, emphasizing the role of frontline professionals in co-creating value and enabling innovation. If professionals are not committed to the technology, full adoption and integration into practice are unlikely to occur. This lens allows for the identification of early drivers of innovation in service delivery. While some studies have examined healthcare professionals' perspectives (e.g. Scipion et al., 2025; Antonacci et al., 2023; Lambert et al., 2023; Hamline et al., 2024; Gagnon et al., 2016), they do not approach the problem from a service management perspective. Published in medical journals, they overlook insights from service management literature on co-creation, patient experience, employee engagement, and value co-production.

Second, this study explores professionals' perceptions of a pre-commercialized device. Investigating pre-market attitudes captures expectations before usage experiences shape opinions. This reflects service innovation logic, recognizing the importance of understanding actors' sense-making in the early stages of technological change. Timing provides insight into how service providers interpret innovation opportunities before routinization.

Third, the analysis adopts a service management and innovation perspective. While studies have examined professionals' attitudes toward technology adoption, few have used a service-oriented theoretical lens. This study investigates the conditions that shape adoption decisions within the service system. The configurational reasoning through QCA extends the UTAUT framework by revealing how multiple factors jointly influence adoption.

Finally, focusing on self-employed professionals minimizes the influence of organizational factors, such as hospital policies or infrastructural limitations. This enables the examination of technology adoption as an autonomous service management decision, which is less affected by organizational hierarchies and more reflective of individual professional judgment.

This study bridges technology adoption research with the service management and innovation domains. By combining configurational QCA analysis with professionals' pre-market evaluations of new technology, this study provides a system-oriented understanding of how innovation unfolds in professional service contexts, moving beyond UTAUT-based models.

As in other professional service contexts (von Nordenflycht, 2010), technology adoption in healthcare cannot be understood merely as a response to the increasing availability of digital tools, but rather as a process shaped by professional identity, service logics, and contextual constraints. Although the digital transformation of healthcare is now well established, the present study does not aim to confirm or challenge this trend. Instead, it focuses on the drivers of technology adoption and the specificities that characterize adoption decisions in professional healthcare settings.

Extant research highlights that digital transformation in healthcare entails a fundamental redesign of service encounters, redistributing roles and value creation activities across the service system (Patrício et al., 2020). However, adoption decisions in healthcare differ markedly from those observed in consumer or purely organizational contexts. In healthcare services, value is co-created through professional–patient interactions, meaning that digital tools reconfigure not only workflows but also professional responsibility, autonomy, and accountability (Løwendahl, 2005, pp. 20–24; von Nordenflycht, 2010). Consequently, professionals must balance client value creation with organizational efficiency and professional legitimacy when considering whether to adopt new technologies (Zainuddin et al., 2016).

The literature on technology acceptance, including TAM and UTAUT-based models, identifies key adoption drivers such as perceived usefulness, ease of use, facilitating conditions, and affective responses. However, in healthcare contexts, these drivers are embedded in a socio-professional environment characterized by regulatory requirements, ethical obligations, and strong professional norms (Makarem and Al-Amin, 2014). Consequently, adoption is rarely an individual-level decision; rather, it is shaped by professional communities, reimbursement structures, and institutional arrangements.

The distinction between professionals and users is particularly salient in healthcare services. Unlike many digital services where users are also the primary adopters, healthcare technologies are often adopted by professionals but used in interaction with patients, who may have limited agency in the adoption decision. Eysenbach's (2001) conceptualization of eHealth, encompassing business-to-consumer (B2C), business-to-business (B2B), and consumer-to-consumer (C2C) interactions, underscores how adoption depends on coordination across multiple actors and levels of the healthcare system. From a service innovation perspective, this implies that usability extends beyond interface design to include governance structures, workflow integration, and professional legitimacy (Dagger et al., 2007).

Within this context, eHealth applications, ranging from telemedicine to digitally supported rehabilitation, illustrate how adoption decisions are shaped by clinical goals and service delivery arrangements (Dodds et al., 2022). For instance, remote rehabilitation for balance disorders shows how digital tools can support the continuity of care while preserving clinical supervision and professional judgment (Meldrum et al., 2022; Pavlou et al., 2012; Zhang et al., 2022). Such applications are often perceived by professionals as strengthening patient relationships across time and space, rather than merely increasing efficiency (Guitton, 2021; Huang et al., 2022).

Overall, adoption in healthcare should be understood as a multi-level and context-dependent process, in which individual perceptions interact with professional norms and ecosystem-level arrangements. This socio-professional complexity makes healthcare a particularly relevant context for examining technology adoption beyond net-effect models, and for exploring how different configurations of adoption drivers lead to the acceptance of digital tools in professional service settings.

Service innovation adoption in professional contexts shows characteristics that variance-based methods cannot capture, particularly equifinal pathways and configurational causation in how providers navigate professional, economic, and relational demands. fsQCA enables identifying alternative service innovation pathways, addressing calls for methods that acknowledge causal complexity in service systems. To study different drivers and combinations, this study applied fsQCA, which links to social complexity. Social complexity involves processes marked by case heterogeneity and variance, bound by “causal recipes”, relating to generational heterogeneity. QCA is useful in revealing social complexity (Byrne and Ragin, 2009; Pappas et al., 2019; Kraus et al., 2018; Ragin et al., 2003; Sager and Andereggen, 2012) and recognizes situations under various conditions (Olsen, 2014). We chose fsQCA because service innovation adoption in healthcare contexts shows causal complexity, professional heterogeneity, and non-linear relationships among organizational, technological, and relational factors (Fiss, 2011; Woodside, 2014). Unlike variance-based methods that assume symmetry, fsQCA uncovers multiple pathways to high adoption, which is essential for professional service contexts. This choice aligns with our objective of identifying how different conditions jointly enable eHealth adoption and answers calls to adopt configurational approaches (Ordanini et al., 2014). FsQCA allows us to distinguish necessary from sufficient conditions to determine how factors must combine to cross the adoption threshold (Ragin et al., 2003; Cooper and Glaesser, 2016). By embracing causal asymmetry, we can model pathways leading to high adoption intentions and examine whether different configurations explain low intentions, which regression techniques cannot capture (Fiss, 2011; Woodside, 2014).

While fsQCA typically does not test conventional hypotheses, we grounded our factor selection in established theory on eHealth, professional service firms, and service innovation (e.g. Eysenbach, 2001; von Nordenflycht, 2010; Løwendahl, 2005, pp. 32–51), while remaining open to discovering novel configurations (Guerola-Navarro et al., 2021; Pappas and Woodside, 2021). FsQCA is tailored to explicating multiple configurational pathways through which professionals intend to adopt eHealth services (Sweeney et al., 2023). It offers an alternative to additive variable-oriented models (e.g. Gerrits and Pagliarin, 2021; Ragin, 2000), focusing on variation within diverse examples and common patterns. Few cross-case generalizations or “causal recipes” can be established, reflecting the link between case heterogeneity and social complexity (Ragin, 2000; Rihoux and Lobe, 2009, pp. 222–242). QCA accounts for complexity through an ontological understanding articulated epistemologically in its processes and analyses.

The fsQCA methodology identifies necessary and sufficient conditions enabling outcomes (Ragin et al., 2003). A state is sufficient when its presence produces an outcome, while multicausality occurs when condition combinations explain outcomes. In set theory, cases with conditions form a subset of those with outcomes, while necessary conditions are present whenever outcomes occur (Cooper and Glaesser, 2016).

We used a platform enabling remote digital administration of exercises that consumers perform at home under supervision of an ENT doctor (Asad et al., 2021). Administration occurs through a laptop platform connected wirelessly with a VR headset loaned by the doctor.

Our data (n = 315) included ENT specialist doctors, self-employed professionals, and balance disorder experts. We recruited 115 participants at the Annual National Congress of the Italian Association of Self-Employed Otolaryngologists in Trento (Italy) in September 2021. An additional 200 ENTs were recruited during 23 local meetings across Italian cities. Data collection occurred as Italy emerged from COVID-19 lockdowns, a timing that influenced professionals' perspectives on remote care delivery. Table 1 describes the participants' demographics and experiences.

To ensure consistency, all meetings with ENTs followed a standardized procedure. The VR system for remote vestibular rehabilitation exercises was installed on-site and connected to technicians remotely. An engineer explained the system's functionality, after which participants tested the device for 30 min and posed questions. The participants then completed a digital questionnaire. To mitigate common method biases, we adhered to the recommendations of Podsakoff et al. (2003), ensuring anonymity, emphasizing no right or wrong answers, and randomizing survey items for all respondents.

A validation and discussion of questionnaire aim, and results was conducted using focus groups to ensure proper understanding of overall aim of the study and to provide comments and integration to the survey questions. Focus groups were integrated into the study formulation, where qualitative methodology was essential post doc analysis tool for valid result interpretation.

This approach is consistent with previous research that has adopted focus groups for a similar post hoc validation role (e.g. Ng et al., 2024; Cerela-Boltunova et al., 2025; Archambault et al., 2014).

The focus groups validated and integrated the questionnaire's understanding and effectiveness. Although the questionnaire defined each construct and related items (Performance Expectancy, Effort Expectancy, Social Influence, Facilitating Conditions, and Behavioral Intentions), healthcare professionals might have limited familiarity with these concepts.

The focus groups served as a complementary qualitative step to ensure understanding of the questionnaire and to discuss the results with participants. This significantly aided the interpretation of the fsQCA analysis. Participants were divided into small groups, as shown in Table 2, for structured discussions on the survey topic and related dimensions. The focus groups provided additional validation, ensuring the consistent interpretation of key constructs and reinforcing the quantitative findings.

Our questionnaire included drivers and dimensions of technology adoption. We relied on the UTAUT2 model, proven effective in business-to-business markets (Duarte and Pinho, 2019; Alazab et al., 2020; Dutta and Shivani, 2020; Haikal et al., 2022). The items originated from Venkatesh et al. (2003, 2012), with subsequent research confirming reliability.

Performance Expectancy (EXPe) is defined as an individual's belief that using a system will assist in career advancement (Kim and Malhotra, 2005; Limayem et al., 2007). Effort Expectancy (EFF) refers to the ease of system use (Davis et al., 1989; Moore and Benbasat, 1991; Thompson et al., 1991). Social Influence (SOCI) is an individual's perception that others believe they should use the new system (Thompson et al., 1991; Venkatesh and Davis, 2000). Facilitating Conditions (FAC) represent the belief that organizational and technological infrastructure exists to support system use (Taylor and Todd, 1995). Hedonic Motivation (HED) is the enjoyment derived from using a system (Brown and Venkatesh, 2005; van der Heijden, 2004). Price Value (PRICE) is the trade-off between an app's advantages and usage cost (Dodds et al., 1991). When benefits exceed costs, Price Value positively influences intention (Dodds et al., 1991; Zeithaml, 1988). Anxiety (ANX) refers to emotional reactions when adopting behavior (Compeau and Higgins, 1995) and was inserted as a potential barrier for our behavioral intention within the UTAUT model for two reasons. First, compared to consumer technology adoption, the level of resistance in the context under investigation may be higher; therefore, we aimed to examine whether this resistance is associated with an emotional state such as anxiety. Second, our choice is supported by prior literature that has incorporated the anxiety construct within the UTAUT framework (as a few examples of studies adopting UTAUT or UTAUT2 that incorporate anxiety, see Chen et al., 2024; Joshi et al., 2025; Pratama et al., 2025; Huang et al., 2025, in the services field). Behavioral Intention (BI) is the subjective probability of technology adoption (Davis et al., 1989; Moore and Benbasat, 1991; Sheppard et al., 1988).

 Appendix lists these constructs and their measurement characteristics.

The estimation of fsQCA comprises three crucial steps. First, the calibration process converts values between 0 and 1, indicating degree of membership in a defined category: 1 indicates full membership, 0 indicates full non-membership, and 0.5 means neither in nor out (the crossover point) (Schneider and Wagemann, 2010a; Schneider, 2018). The crossover point was calculated by observing distribution and median scores of each attribute, consistent with Greckhamer et al. (2018). To simplify analysis, the original Likert scale variables were transformed into three categories: 0.95 (for 6 and 7), 0.5 (for 4 and 5), and 0.05 (for 1, 2, and 3). For BI outcomes, we used the same direct calibration method.

Second, a truth table was created to explain the professionals' intention to adopt the eHealth service and the causal factors necessary for the outcome. The truth table treats each case as a configuration, with cases sharing configurations considered the same type. Each row lists all possible 3k combinations of causal conditions (k being the number of conditions; here 37, equaling 2,187 combinations; Ragin, 2009). The truth table includes configurations meeting consistency criteria, with a minimum threshold of 0.9, higher than Ordanini et al.’s (2014) suggested 0.75.

Finally, the software computes solution coverage, indicating cases satisfying conditions. Coverage represents “empirical importance” (Schneider and Wagemann, 2010b, p. 407). Following fsQCA best practices, we used a consistency cut-off of 0.80 (Ferguson et al., 2017; Ragin, 2009), minimum case cut-off of 2 (Ordanini et al., 2014) and reported intermediate and complex solutions (Pappas and Woodside, 2021).

Our fsQCA analysis required transforming Likert-scale responses into fuzzy set membership scores through direct calibration.

As sets are expressed in binary form (presence/absence of attributes), and our variables are not naturally dichotomous; we transformed construct measures into fuzzy-set membership scores, calibrating measures by specifying three qualitative anchors: the threshold for full membership in a set (i.e. value 0.95), the threshold for full non-membership in a set (i.e. value 0.05), and the crossover point (i.e. value 0.5) (Ragin and Fiss, 2008). This is in line with previous best practices (Pappas and Woodside, 2021; Schneider and Wagemann, 2012), which adopted three-point calibration approach that helps not only to capture the presence or absence of a phenomenon but also the crossover point.

As we needed to manage multiple-item measures, scale items were combined into an average score (Leischnig and Kasper-Brauer, 2015). The endpoints and midpoints of the 7-point Likert scales served as the three qualitative anchors for the calibration of full membership (value 6–7), full non-membership (values 1–3), and the crossover point (value 4–5) (Russo et al., 2016).

This transformation was justified because meaningful distinctions in professional adoption contexts typically occur at these thresholds, as established by Greckhamer et al. (2018). This approach ensures adequate case distribution while maintaining interpretive clarity. Sensitivity analyses with alternative calibration thresholds confirmed that our core configurational relationships remained stable, indicating robust set-theoretic patterns. This is in line with common methods employed to ensure the robustness of QCA results, including adjusting the calibration threshold, changing the consistency threshold, adding or removing cases, and changing the frequency threshold (Pappas and Woodside, 2021).

Following Ragin's (2009) two-stage analytical framework, we first conducted a necessity analysis to identify the conditions that must be present for high adoption intention and then examined sufficient configurations.

Following established thresholds (Ragin, 2009), conditions with consistency ≥0.90 and coverage ≥0.5, meet the necessity criteria. Table 3 presents the necessity analysis results.

Four conditions meet the necessity thresholds:

1. Performance Expectancy (EXPe): consistency = 0.94, coverage = 0.76

2. Hedonic Motivation (HED): consistency = 0.92, coverage = 0.71

3. Price Value (PRICE): consistency = 0.91, coverage = 0.73

4. Social Influence (SOCI): consistency = 0.90, coverage = 0.68

These four conditions are present in ALL cases of high adoption intention. No professional in our sample expressed high adoption intention without perceiving performance gains, experiencing intrinsic satisfaction, seeing economic value, and feeling social validation. This simultaneous necessity distinguishes professional service innovation from contexts where single factors suffice.

Three conditions did NOT meet necessity thresholds:

1. Effort Expectancy (EFF): consistency = 0.78 (below 0.90 threshold)

2. Facilitating Conditions (FAC): consistency = 0.81 (below 0.90 threshold)

3. Low Anxiety (∼ANX): consistency = 0.85 (below 0.90 threshold)

While these conditions appear in many high-intention cases, they are not indispensable. This finding is theoretically significant: ease of use and infrastructure support help but don't guarantee adoption when necessary conditions are met.

We constructed a truth table with all possible combinations of the seven conditions (Ragin, 2009), applied consistency threshold of 0.80 and minimum case threshold of 2, and analyzed both intermediate and complex solutions Table 4.

The analysis identified ONE sufficient configuration:

Consistency = 0.97 (well above 0.80 threshold)

Raw coverage = 0.63 (explains 63% of high adoption cases)

This configuration indicates that when professionals have Effort Expectancy (perceive ease of use), Facilitating Conditions (adequate infrastructure support), and low Anxiety (comfort with technology) AND these combine with the four necessary conditions, high adoption intention results.

Our fsQCA reveals a hierarchical causal architecture:

• TIER 1 – Necessary Conditions (Universal Requirements): All four must be present simultaneously: Performance Expectancy (EXPe), Hedonic Motivation (HED), Price Value (PRICE), and Social Influence (SOCI).

• TIER 2 – Sufficient Configuration (When Added to Necessary Conditions): The combination of Effort Expectancy (EFF), Facilitating Conditions (FAC), and Low Anxiety (∼ANX).

In set-theoretic notation, high adoption occurs when:

Where:

1. The bracketed necessary conditions are always present (in 100% of high adoption cases)

2. The sufficient configuration, when combined with necessary conditions, produces high intention in 63% of cases

3. The asterisk (*) denotes logical and

4. The tilde (∼) denotes absence/negation

This two-tier structure implies that professional service innovation needs foundational prerequisites (necessary conditions for motivational, economic, and social dimensions) and enabling circumstances (sufficient configuration addressing implementation barriers). Technology vendors addressing only ease of use and support will fail if they neglect necessary motivational and economic foundations.

These findings reflect an interplay between expectancy-value mechanisms and self-determination processes. Performance Expectancy aligns with instrumental beliefs in expectancy theory, as professionals assess technology's utility for maintaining clinical performance. Hedonic Motivation reflects intrinsic motivation, where enjoyment and satisfaction sustain engagement beyond utilitarian value, consistent with self-determination theory. Price Value captures economic rationality and cost-benefit evaluation, particularly relevant for self-employed professionals bearing investment costs. Social Influence represents normative pressures through which peer and patient expectations shape intentions, reflecting Social Cognitive Theory principles.

The combined configuration (HED * EXPe * PRICE * SOCI * EFF * FAC * ∼ANX) indicates that adoption requires cognitive evaluations and affective assurance. Low Anxiety enables learning when accompanied by Facilitating Conditions and perceived ease of use. These patterns show complementarity between perceived control (EFF, FAC) and emotional regulation (∼ANX) in supporting technology adoption.

Contextual factors explain these mechanisms. The Italian healthcare system positions ENT specialists as semi-autonomous decision-makers balancing clinical and economic factors, making EXPe and PRICE necessary conditions. Post-pandemic shifts amplify SOCI as peer behavior and patient demand normalize hybrid services. The coexistence of Hedonic Motivation and Performance Expectancy shows that adoption is both instrumental and identity-affirming, as professionals gain satisfaction from mastering innovative tools.

These mechanisms integrate Expectancy Theory, Social Cognitive Theory, and Self-Determination Theory within a configurational service innovation framework.

Our findings extend von Nordenflycht's (2010) PSF theory by showing how PSF traits affect service innovation adoption. Von Nordenflycht (2010) identifies three key characteristics: knowledge intensity, low capital intensity, and professionalized workforce. Our results highlight how each influences technology adoption.

Hedonic Motivation (consistency = 0.92) reflects PSFs' knowledge-intensive nature. Unlike routine services, professional service delivery is tied to expertise and identity. Healthcare professionals need intrinsic satisfaction from technology as it becomes part of their expertise demonstration. Unsatisfying technology threatens professional identity regardless of utility. This makes Hedonic Motivation crucial in professional contexts but often peripheral in consumer technology or organizational IT, where usage can be mandated.

Price Value (consistency = 0.91) reflects PSFs' low capital intensity. Self-employed healthcare professionals lack organizational buffers for technology costs. Unlike hospital-employed physicians, they make direct cost-benefit calculations impacting personal income. Each euro in eHealth reduces earnings unless offset by efficiency. This explains why Price Value is necessary in our sample but often excluded from studies where users don't bear costs.

Social Influence (consistency = 0.90) reflects the professionalized workforce. Professional identity relies on peer recognition. Technology adoption is a professional statement visible to colleagues and patients. Adopting technologies outside peer norms risks appearing cutting-edge or reckless depending on normalization. This social mechanism differs from consumer technology (dominated by fashion) and organizational IT (where mandates override peer influence).

Performance Expectancy (consistency = 0.94) is necessary in all contexts but varies by PSF. For self-employed professionals, “performance” includes clinical effectiveness, practice efficiency, patient satisfaction, and income generation. This differs from consumer contexts (personal benefit) and organizational contexts (employer-defined metrics). The configuration (EFF * FAC * ∼ANX) addresses implementation barriers, enabling adoption when necessary conditions are met. Motivated professionals, with reduced effort, infrastructure, and minimized anxiety, are inclined to adopt. However, focusing only on these factors without motivation, economic viability, and social legitimacy leads to failure, explaining disappointing technology rollouts focused solely on training and IT support. These findings advance PSF theory by showing von Nordenflycht's (2010) characteristics in service innovation adoption. Our work extends PSF theory from industry characteristics to innovation adoption mechanisms within PSFs, complementing macro-level theory by showing industry characteristics in individual decision-making.

Our findings extend rather than replicate UTAUT by revealing necessity, sufficiency, and configurational patterns that net-effects models cannot detect. Table 5 summarizes key distinctions.

UTAUT and UTAUT2 (Venkatesh et al., 2003, 2012) show that Performance Expectancy, Effort Expectancy, Social Influence, Facilitating Conditions, hedonic motivation, and Price Value predict Behavioral Intention. These models determine adoption factors and their strength. In healthcare contexts, Performance Expectancy shows strongest effect (β ≈ 0.30–0.45), followed by Effort Expectancy (β ≈ 0.20–0.35), with other factors showing weaker effects.

Our fsQCA analysis reveals three insights beyond UTAUT's net-effects logic:

The distinction between variance-based and configurational approaches extends beyond statistical technique to causal interpretation. Traditional UTAUT studies assume linear, additive, and symmetric causation, where each factor contributes independently to the outcome, and factors can partially compensate for one another. Under this logic, a healthcare professional scoring 7/7 on Performance Expectancy but only 4/7 on Hedonic Motivation would be predicted to have moderate-to-high adoption intention, as the strong Performance Expectancy score compensates for moderate Hedonic Motivation.

Our configurational analysis revealed that this prediction was incorrect. Since Hedonic Motivation is a necessary condition (consistency = 0.92), scores below the threshold preclude high adoption intention regardless of Performance Expectancy levels. This represents non-compensatory causation: necessary conditions cannot be offset by high scores on other factors. As Table 5 demonstrates, all four necessary conditions (EXPe, HED, PRICE, SOCI) must reach threshold levels simultaneously—a structural requirement that variance-based models cannot detect.

This distinction has profound implications for practice. While UTAUT-based recommendations typically focus on “maximizing the strongest factor” (usually Performance Expectancy given its highest beta weight), our configurational insight suggests a different strategy: “ensure all necessary conditions exceed thresholds—address the weakest link, not just the strongest opportunity”. For healthcare technology providers, this means that improving ease of use (EFF) or infrastructure support (FAC) will not drive adoption if any necessary condition (EXPe, HED, PRICE, SOCI) remains below the threshold. Resources must first ensure alignment across all four dimensions.

• Contribution 1: Necessity Information

We identify four necessary conditions that must exist simultaneously. UTAUT shows these factors have positive effects but does not distinguish necessary from helpful factors. The distinction matters theoretically and practically. Professional service innovation requires multi-dimensional alignment rather than strong performance on any dimension, reflecting high-stakes work. Technology providers cannot compensate for weakness in one necessary condition by strengthening others. High performance (EXPe) doesn't overcome poor economic value (low PRICE) or lack of peer adoption (low SOCI).

• Contribution 2: Sufficiency Configuration

We identify a configuration (EFF * FAC * ∼ANX) that produces high intention with 0.97 consistency. UTAUT's additive logic assumes factors increase adoption linearly. Our configurational logic reveals: (1) multiple adoption pathways exist (Equifinality, with one dominant configuration), (2) factors matter in combination (Conjunctural Causation), and (3) the absence of anxiety matters, while its presence doesn't drive adoption (Asymmetry).

• Contribution 3: Professional Service Specificity

We reveal a pattern in professional services differing from UTAUT applications in consumer and organizational settings. In consumer technology, performance and Effort Expectancy dominate, with Hedonic Motivation for leisure, Price Value for payments, and weak social effects. Organizational IT emphasizes Performance Expectancy and Facilitating Conditions, with hierarchical Social Influence. For professional services, Performance Expectancy, Hedonic Motivation, Price Value, and Social Influence are all essential, while effort and Facilitating Conditions enable adoption. This pattern reflects autonomous decision-making, direct costs, identity-relevant work, and peer-validated norms.

To verify findings, focus groups helped understand adoption motivations. From seven variables, four were necessary and sufficient (EXPe, HED, PRICE, and SOCI) for high Behavioral Intention (BI) values. As noted in the methodology, data collection timing influenced professionals' heightened awareness of remote care needs. As Figueroa et al. (2019) explained, technological innovation responds to medicine's essential needs. EXPe was necessary as self-employed ENTs without fixed salaries needed technology for remote practice. Focus groups confirmed the necessity of HED, connecting technology adoption to professional satisfaction. PRICE is necessary as health professionals resist eHealth adoption despite cost-effectiveness (Virlée et al., 2020; Patel et al., 2022). The sample showed concern for investment returns during lockdowns (Chauhan et al., 2022).

Although eHealth was pivotal during COVID-19 (Huang et al., 2022), SOCI leads to high BI values when combined with other variables. External pressures changed attitudes toward technology, shifting to hybrid systems. Social Influences, including patient expectations (Jiang, 2022), changed health professionals' behavior.

To understand how the eHealth setting shapes fsQCA results, we examined mechanisms. Professional autonomy made Performance Expectancy (EXPe) salient and kept Price Value (PRICE) central to adoption. The Italian setting's mix of private and public practice reinforced Social Influence (SOCI). Patient expectations explain SOCI's necessity. Post-pandemic pressures made workflow changes acceptable, visible where Effort Expectancy (EFF) and Facilitating Conditions (FAC) are present and technology anxiety (ANX) is low. The specialty services' features explain the EFF and FAC pairing with low ANX. These contextual forces appear in the solution structure, expressed as HED * EXPe * PRICE * SOCI * EFF * FAC * ∼ANX, where “∼ANX” denotes anxiety absence.

While EFF and FAC were positive, ANX required absence rather than presence. This is typical due to digital skill gaps (e.g. Di Giacomo et al., 2019), leading to studies of “technophobia” (Compeau et al., 1999; Osiceanu, 2015). The sample did not show a high initial condition value.

QCA results show technology adoption requires effort in adapting to remote patient relationships (Wu et al., 2020). EFF data aligns with low anxiety levels, with respondents showing comfort with technology. Research suggests (Alshammari et al., 2022; Zoghlami and Ben Rached, 2024) that healthcare professionals' adoption decisions are not strongly linked to digitalization fears. Adoption requires effort, without major concerns about training or support.

These patterns reflect ecosystems where adoption decisions combine professional identity, client value, and peer expectations. The necessity of Hedonic Motivation with Performance Expectancy indicates that innovation needs satisfaction beyond utilitarian value. Our focus on Behavioral Intention aligns with research where intention predicts usage (Venkatesh et al., 2012). For ENT specialists, intention matches adoption decisions due to service control.

These forces show how institutional logics shape provider adoption intentions, supporting service logic's emphasis on context-dependent value creation. These relationships indicate service innovation emerges through resource integration rather than substitution, extending understanding of digital transformation.

This study advances professional service innovation theory by revealing the configurational structure of provider adoption decisions. Using fsQCA with 315 ENT specialists in Italy, we show that digital healthcare service adoption requires a two-tier structure: four necessary conditions that must be present (Performance Expectancy, Hedonic Motivation, Price Value, Social Influence), combined with sufficient conditions enabling adoption (Effort Expectancy, Facilitating Conditions, low Anxiety).

This configuration distinguishes professional service innovation from consumer technology adoption and organizational IT implementation. Unlike contexts where single factors drive adoption, professional service contexts require multi-dimensional alignment: knowledge intensity makes intrinsic satisfaction necessary, low capital intensity makes economic viability necessary, and professionalized workforce makes peer validation necessary.

Our findings integrate Expectancy Theory, Self-Determination Theory, and Social Cognitive Theory, showing these mechanisms operate simultaneously in professional adoption decisions. As detailed in Section 4.5, our configurational findings extend technology adoption models by revealing how professional service characteristics shape individual adoption patterns.

Using fsQCA, we show adoption results from specific factor combinations. The four necessary conditions (Section 4.1) combine with complementary conditions (Effort Expectancy, Facilitating Conditions, lower Anxiety) to enable high Behavioral Intention. In our Italian ENT context, respondents described shifting to hybrid care, with peer expectations normalizing use, explaining SOCI's necessity alongside EXPe, HED, and PRICE. The sufficiency solution shows professionals invest in learning when basic enablers exist and Anxiety is low. Focus group accounts align with configurations where EFF and FAC are present and ANX is muted. Economic evaluation is crucial, with PRICE present in every high-intention pathway.

Methodologically, these findings demonstrate fsQCA's value in professional settings: showing equifinality and asymmetry with substantial coverage. This approach separated necessary elements from complementary ones, creating a clearer design map. Adoption emerges when professionals expect performance gains and satisfaction, see credible price-value, align with peer norms, and implementation reduces effort while minimizing Anxiety.

These configurational insights provide managers with strategic levers for service transformation.

Our study makes three primary theoretical contributions, advancing the understanding of professional service innovation adoption beyond the existing technology acceptance and service management literature.

First, our configurational analysis (Section 4.1-4.3) reveals that professional service innovation adoption requires simultaneous alignment across four dimensions: cognitive (Performance Expectancy), affective (Hedonic Motivation), economic (Price Value), and social (Social Influence). This distinguishes professional service contexts from consumer technology adoption and organizational IT implementation. By adopting a configurational perspective, our study advances service innovation theory by demonstrating that technology adoption by providers differs fundamentally from patient adoption during healthcare service transformation (see, for example, Aggarwal et al. (2017) and Hategan et al. (2019). This research shows how digital transformation reshapes professional service ecosystems. By focusing on healthcare professionals, this study demonstrates that provider adoption is critical for digital health innovation. Service management literature shows that digital transformation requires mechanisms that align professional motivation and trust (Fusco et al., 2023; Borges do Nascimento et al., 2023). Our findings show that successful technology-enabled services depend on configurations that combine performance with emotional and social legitimacy.

These results suggest that policymakers should view professional adoption as a lever for equitable access. When professionals view innovations as rewarding, they extend services to underserved populations (Danaher et al., 2024; Lambert et al., 2023). These understanding bridges individual motivation and system outcomes.

Second, we integrated psychological theories–Expectancy Theory, Self-Determination Theory, and Social Cognitive Theory–within a configurational framework.

Therefore, our approach goes beyond the mere application of the UTAUT. Rather than relying on a single theoretical lens, we integrate these frameworks from managerial and configurational perspectives.

Traditional models apply these theories additively; we reveal that they operate conjuncturally in professional contexts. Performance Expectancy cannot compensate for the absence of Hedonic Motivation, nor can it overcome weak social validation. This integration advances service innovation research by demonstrating that provider adoption involves simultaneous psychological processes.

Third, we contribute methodologically by demonstrating the value of fsQCA for professional service research. By distinguishing necessary from sufficient conditions and revealing configurational patterns, fsQCA complements variance-based approaches in technology adoption research.

Therefore, our approach goes beyond the mere application of fsQCA. We do not claim that the use of QCA per se is novel; instead, we argue that the configurational results obtained are more robust because they embrace the inherent complexity of the healthcare service context. This approach allowed us to link the constructs to one another, highlight their differential importance, and offer a more comprehensive and nuanced version of the adoption model.

Our findings show that configurational methods reveal adoption mechanisms that net-effects models miss, particularly where causal complexity characterizes decision processes.

The configurational approach revealed equifinal paths and causal asymmetry, showing joint necessity of Performance Expectancy, Hedonic Motivation, Price Value, and Social Influence in technology adoption. Results highlight how adoption requires cognitive evaluation and affective assurance through Hedonic Motivation and Social Influence, reinforcing debates about psychological complementarity between perceived control and emotional regulation. Our findings answer to a call for research to inform healthcare digitization approaches (Patricio et al., 2020). By adopting emerging technologies like augmented and virtual reality, our study provides inputs for evaluating human experiences.

Our findings provide evidence-based guidance that differs from generic technology adoption strategies. Traditional approaches assume compensatory logic, where strong performance compensates for weak ease of use. Our configurational findings reveal non-compensatory logic in professional service contexts, showing all necessary conditions must be addressed simultaneously.

For technology vendors and service designers, generic value propositions emphasizing single benefits will fail with professional service providers. Design interventions must address all four necessary conditions simultaneously. Performance must be demonstrated through local clinical effectiveness data. Hedonic value must enable professional mastery and expertise demonstration. Economic viability requires transparent ROI calculations with reversible risk structures. Social legitimacy must be established through specialty-specific peer validation.

Our two-tier causal structure suggests strategic implementation sequencing. The first phase establishes necessary conditions through pilots demonstrating local performance outcomes, technology supporting professional satisfaction, pricing aligned with professional economics, and peer networks providing specialty community legitimacy. The second phase reduces implementation barriers through comprehensive training, technical infrastructure, and adoption support. Attempting the second phase without completing the first wastes resources.

For healthcare leaders and policymakers, necessary conditions require system-level interventions. Performance Expectancy demands alignment of metrics with digital delivery models. Hedonic Motivation requires including professionals in technology selection to preserve autonomy. Price Value necessitates subsidies or reimbursement adjustments for self-employed professionals. Social Influence requires engaging professional associations early, as specialty society endorsement matters more than generic marketing campaigns.

Once system-level necessary conditions are established, the sufficient configuration requires practice-level support through training, IT assistance, and graduated implementation that builds confidence through early successes rather than comprehensive deployment.

Organizations can apply a diagnostic framework by evaluating all necessary conditions. If any necessary condition is weak, adoption will fail regardless of other strengths, requiring targeted intervention before proceeding. If all necessary conditions are strong but adoption lags, focus should shift to enhancing training, support, and Anxiety reduction. If both necessary and sufficient elements are present but adoption remains low, investigation should examine contextual barriers like patient resistance or competing priorities that constrain decision-making.

This diagnostic logic differs from traditional technology adoption approaches that prioritize addressing the weakest factor. In professional service contexts, the weakest necessary condition creates an absolute constraint. Strengthening strong factors yields no benefit when any necessary condition remains inadequate. This non-compensatory logic reflects the high-stakes nature of professional service innovation, where partial solutions fail because professional, economic, and social dimensions must align for adoption.

Clinicians form adoption intentions when they see how a service sustains clinical activity and income, when it feels satisfying, when price matches value, and when peers and patients expect its use. Vendors should redesign their service value proposition accordingly. Proposals should include ROI worksheets from local data, a time-to-first-benefit plan with milestones, and a pricing model with reversible risk. Marketing materials should feature peer champions from the same specialty and region, and patient stories that set new norms.

Practice leaders should stage adoption as a managed service modification rather than a voluntary add-on, given that successful configurations include supportive effort expectations and low Anxiety. A three-month plan should assign a super-user, schedule rehearsals, and allocate protected time. A readiness checklist should verify bandwidth, devices, templates, and procedures, with leaders monitoring Anxiety and providing support during initial cases. National societies should curate peer case exemplars, create micro-credentials for remote care, and maintain a registry of accredited tools. For payers and policymakers, Price Value is crucial for self-employed clinicians. Payers can pilot per-episode fees with interoperability grants, while regulators can publish compliance kits to strengthen Facilitating Conditions.

Implementation should include a process map of assessment and follow-up, an integration checklist, and a playbook for technical issues. Success should be tracked through three indicators: eligible patients offered the service, offered patients enrolled, and median time from referral to first remote session. These moves align with the finding that intention is highest with supportive conditions.

Rather than calling for a generalized culture of change, we specified steps linking directly to patient and system outcomes. Providers and vendors should target indications where remote interaction is clinically appropriate and care continuity is threatened by travel or capacity constraints, then report two patient-level outcomes, such as program completion and symptom improvement, and two system-level outcomes, such as avoided visits and time to assessment. Public agencies can support these pilots with small interoperability grants and outcome dashboards, which creates visible social proof while building Facilitating Conditions that our results show drive successful configurations.

Our empirical evidence is drawn from Italian ENT specialists, many of whom operate as self-employed professionals in market-oriented outpatient settings. In this context, price (PRICE) and perceived adoption effort (EXPe) are particularly salient because clinicians directly bear acquisition, maintenance, and workflow-adjustment costs. By contrast, in salaried settings, centralized technology budgets and dedicated IT support may buffer these trade-offs, shifting the relative importance toward Facilitating Conditions (FC) and Social Influence (SI). More broadly, system-level factors moderate the identified configurational pathways: payment models shape PRICE and Performance Expectancy (PE); digital maturity influences EXPe and perceived risk; telehealth regulations affect trust, FC, and PE; and cultural norms regarding professional autonomy and privacy shape SI and risk perceptions. These contextual differences limit the straightforward generalizability of the Italian configurations.

As shown in Section 4.5 and Table 5, the configurational approach reveals multiple equifinal pathways and necessity–sufficiency relationships that extend UTAUT-based frameworks (Fiss, 2011; Ragin and Fiss, 2008). Future research could explicitly compare employment models to test the role of price value in salaried environments, extend the analysis to other healthcare services, and adopt longitudinal designs to examine shifts in necessity conditions over time. Mixed-methods approaches that combine qualitative interviews with discrete-choice experiments could further quantify trade-offs between pricing and technological features.

Importantly, configurations are not static. As clinicians accumulate experience, reductions in EXPe may alter sufficiency patterns, while peer adoption and evolving patient expectations can strengthen SI. At the same time, vendor competition may lower PRICE while enhancing PE, and health system policies may reshape FC through funding mechanisms and standard-setting. Improvements in digital infrastructure can simultaneously reduce EXPe and increase PE, whereas exogenous shocks may recalibrate perceived risk. Because fsQCA relies on calibrated thresholds, necessity and sufficiency relationships may shift even when underlying constructs remain conceptually stable.

Building on these insights, future studies should examine how self-employed versus salaried professionals weight PRICE, EXPe, and FC across institutional contexts, extend the analysis to settings with strong organizational mandates, and employ cross-national fsQCA to test configurational invariance. Longitudinal designs could track dynamic changes in EXPe, SI, PRICE, PE, and FC, while multilevel models would allow researchers to capture cross-level interactions between individual, organizational, and system-level factors.

Finally, because our sample consists of self-employed professionals, organizational influences such as hospital policies and infrastructural constraints are intentionally minimized. While this enhances insight into market-based professional services, it also constrains transferability to salaried contexts. Comparative studies across countries, employment arrangements, and professional service domains would help establish boundary conditions. Extending this configurational logic to other knowledge-intensive services such as legal, consulting, or engineering firms may further distinguish universal adoption mechanisms from those that are context specific.