Enhancing customer value co-creation and stickiness in social commerce: integrating PLS-SEM and NCA for deeper insights into customer-to-customer dynamics

The study addresses why and how users become "sticky" to social commerce (s-commerce) platforms, focusing on electronic customer-to-customer interaction (eCCI) as a key differentiator from traditional e-commerce. It investigates whether and how eCCI drives customer stickiness via customer value co-creation (functional and hedonic value) and whether self-efficacy conditions these effects. The research highlights gaps: limited empirical work on eCCI in s-commerce, scarce examination of its impact on stickiness as an online behavior, and underexplored links between value co-creation and stickiness in s-commerce. Guided by customer-dominant logic (C-D logic), the study proposes and tests a moderated mediation model where eCCI affects stickiness directly and indirectly via value co-creation, with self-efficacy moderating the indirect path. The work is important for understanding mechanisms of user retention on s-commerce platforms and for informing platform strategies to enhance engagement and competitive advantage.

Customer-to-customer interaction (CCI) is a dynamic exchange of information among customers; its online counterpart, eCCI, is critical in s-commerce where social interaction shapes experience. Drawing on C-D logic, customers are central value creators within their broader consumption ecosystems, co-creating value through interactions, experiences, and emotions. In s-commerce, eCCI facilitates co-creation of two principal value dimensions: functional value (utilitarian benefits like information and problem-solving) and hedonic value (emotional benefits such as enjoyment and happiness). Prior studies indicate these values motivate user behaviors in s-commerce. Stickiness denotes prolonged use and frequent revisits and is a precursor to outcomes like loyalty and purchase intention, driven by perceived value. Self-efficacy, a belief in one's capability to perform actions, influences online sharing and continued platform use; higher self-efficacy can enhance engagement and value creation. The review also situates the study within C-D logic scholarship (e.g., Heinonen et al., Cheung & To, Fang et al.), emphasizing customer-centric value creation across contexts and underscoring the need to examine eCCI's role and value co-creation in s-commerce.

Design: Cross-sectional survey with measurement model and structural model estimation via PLS-SEM, complemented by Necessary Condition Analysis (NCA) for necessity logic. Measures: 5-point Likert scales (1 = strongly disagree, 5 = strongly agree). eCCI: 5 items (Wu & Fang, 2010; Alden et al., 2016). Functional value: Voss et al. (2003). Hedonic value: 7 items (Babin et al., 2007; Kuo & Feng, 2013). Self-efficacy: adapted from McKee et al. (2006). Customer stickiness: 6 items (Lin, 2007; Hsu & Lin, 2016). Items were translated using forward-backward translation; expert panel (two professors, five graduate students) reviewed and refined items. Setting and sample: Users of two Chinese s-commerce platforms—Douyin (TikTok) and WeChat. Inclusion: age 18+ and prior experience with eCCI on these sites. Data collection: August–September 2022 via Wenjuanxing. Pilot: 50 users (35 in person; remainder online). Final sample: 238 valid responses (after removing 42 incomplete); 55.9% male; age 18–27: 44.1%; employed: 41.2%; >4 years s-commerce experience: 36.1%. Data analysis: SmartPLS 4.0 with non-parametric bootstrapping (5000 subsamples). Hierarchical model: customer value co-creation specified as a second-order formative construct with functional and hedonic first-order reflective dimensions. Reliability and validity assessed via loadings, AVE, CR, Cronbach’s alpha; discriminant validity by Fornell–Larcker and HTMT. CMV assessed via common method factor (variance ratio ~74:1; method variance ~0.9%) and full collinearity VIFs (1.343–2.921 < 3.3). Non-response bias tested via extrapolation (t-tests: no significant differences). Mediation tested by bootstrapping (bias-corrected 95% CIs). Moderated mediation tested using Hayes’ PROCESS Model 14 in SPSS (5000 bootstraps). NCA conducted post hoc to identify necessary conditions, including bottleneck analysis.

Measurement model: All constructs showed adequate reliability and validity (e.g., CR = 0.896–0.944; AVE = 0.564–0.736; loadings > 0.6; HTMT < 0.85). CMV and non-response bias not of concern. Structural model (PLS-SEM): - eCCI → Customer value creation: β = 0.414, p < 0.001, 95% CI [0.300, 0.529], f2 = 0.206 (H1 supported). - Customer value creation → Customer stickiness: β = 0.587, p < 0.001, 95% CI [0.492, 0.670], f2 = 0.741 (H2 supported). - eCCI → Customer stickiness: β = 0.305, p < 0.001, 95% CI [0.207, 0.400], f2 = 0.196 (H5 supported). - Indirect effect (mediation): eCCI → Customer value creation → Customer stickiness: indirect effect = 0.243, t = 6.624, 95% CI [0.175, 0.318] (H3 supported; partial mediation). Explained variance (PLS): R2 = 0.171 for customer value creation; R2 = 0.628 for customer stickiness. Moderated mediation (PROCESS Model 14): - Interaction (Customer value creation × Self-efficacy) on stickiness: β = 0.142, p = 0.020. - Conditional indirect effect increases with self-efficacy (at −1 SD: 95% CI [0.472, 0.787]; at mean: 95% CI [0.635, 0.857]; at +1 SD: 95% CI [0.724, 1.000]). - Index of moderated mediation: β = 0.045, 95% CI [0.011, 0.090] (H4 supported). NCA: eCCI is a necessary condition for customer value co-creation (effect size d = 0.095, p < 0.001) and for customer stickiness (d = 0.297, p < 0.001). Bottleneck analysis: to achieve >80% customer value co-creation, eCCI must be ≥ 5.462%; to achieve 80% stickiness, customer value co-creation should be ≥ 42.857%.

Findings validate the theorized C-D logic–based framework in s-commerce: eCCI directly and indirectly (via value co-creation) increases customer stickiness, demonstrating that peer interactions are core mechanisms shaping retained use. eCCI co-creates both functional (information, problem-solving) and hedonic (enjoyment, social pleasure) value, which in turn strengthens stickiness, aligning with prior value-based explanations of online engagement. The moderating role of self-efficacy shows that users’ perceived capability amplifies the value–stickiness linkage and the mediated impact of eCCI, clarifying when and for whom eCCI most effectively translates into persistence on platforms. Integrating PLS-SEM (sufficiency logic) with NCA (necessity logic) provides a richer causal understanding, revealing that while eCCI and value co-creation are sufficient drivers, minimal levels are also necessary to reach high stickiness thresholds. Collectively, the results address the research questions by pinpointing eCCI as a critical antecedent, explicating value co-creation as the mechanism, and identifying self-efficacy as a boundary condition for sustained engagement.

The study advances s-commerce theory and practice by demonstrating that eCCI is a pivotal driver of customer stickiness both directly and through customer value co-creation (functional and hedonic). It extends customer-dominant logic to s-commerce by modeling value co-creation as a second-order construct and by showing that self-efficacy moderates the mediated pathway from eCCI to stickiness. Methodologically, combining PLS-SEM and NCA yields complementary sufficiency and necessity insights, identifying critical thresholds for achieving high value co-creation and stickiness. Practically, platforms should enhance diverse eCCI channels, design for both functional and hedonic value, and incorporate features and guidance that raise users’ self-efficacy (e.g., usability enhancements, tutorials, gamification). Future research can broaden contexts, include additional outcomes (retention, repurchase), examine demographic moderators (e.g., gender), and employ longitudinal and cross-lagged designs to track stickiness over time.

Generalizability is constrained by the Chinese sample; replication in other countries is needed. The model omits other relevant outcomes and predictors (e.g., retention, repurchase intention). Potential demographic moderators such as gender were not tested. The cross-sectional, single-wave design limits causal inference; longitudinal and cross-lagged analyses are recommended to validate temporal dynamics of stickiness.