Beyond Adoption: Unpacking Drivers and Barriers of Customer Mobilization in Digital Logistics Ecosystems A PRISMA Based Systematic Review (2020–2026)

2.1 Review design

This study follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines to ensure transparency and reproducibility. The review was designed as a systematic literature review (SLR), a method well-established in logistics and supply chain management research for synthesising fragmented empirical evidence into coherent theoretical frameworks (Tranfield et al., 2003; Seuring & Müller, 2008).

Research questions were formulated using the SPIDER framework (Setting – Phenomenon of Interest – Design – Evaluation – Research type), which is suited to management and social science research where the phenomenon involves human behaviour within organisational contexts (Cooke et al., 2012). The setting is digital logistics platforms and ecosystems; the phenomenon of interest is customer mobilization, encompassing adoption, engagement, satisfaction, and loyalty; the design includes quantitative, mixed-method, qualitative, expert survey, and structured review studies; the evaluation focuses on drivers, barriers, and emerging trends; and the research type encompasses empirical, conceptual, and structured review contributions. The four research questions were subsequently evaluated against the FINER criteria (Feasible, Interesting, Novel, Ethical, Relevant; Hulley et al., 2007), confirming their appropriateness for systematic investigation.

2.2 Search strategy

A single-database search was conducted in Scopus, selected for its comprehensive coverage of peer-reviewed journals in logistics, supply chain management, and information systems the three fields most central to the phenomenon under review. Scopus is widely used as the primary or sole database in SLRs in the logistics and management domain (Durach et al., 2017; Lim et al., 2021). The limitation of single-database coverage is acknowledged in Section 4.4.

The search was executed on 17 March 2026 using the TITLE-ABS-KEY field. The search string was structured as a three-cluster Boolean query. Cluster 1 captures customer behaviour constructs (“customer mobilization” OR “customer engagement” OR “customer adoption” OR “customer behavior” OR “user behavior” OR “consumer behavior” OR “technology acceptance” OR “platform adoption” OR “customer loyalty” OR “customer satisfaction”). Cluster 2 defines the digital logistics domain (“digital logistics” OR “e-logistics” OR “smart logistics” OR “logistics digitalization” OR “logistics platform” OR “digital freight” OR “on-demand logistics” OR “last-mile delivery” OR “logistics ecosystem” OR “digital supply chain”). Cluster 3 specifies the research focus (“drivers” OR “barriers” OR “challenges” OR “enablers” OR “adoption” OR “trust” OR “perceived usefulness” OR “service quality” OR “digital transformation” OR “technology adoption”). The three clusters were combined using the AND operator, ensuring all retrieved records address customer behaviour, the digital logistics domain, and the research focus simultaneously.

Five automated filters were applied within Scopus: publication year 2020–2026; document type restricted to journal articles; source type restricted to journals; language limited to English; and publication stage set to final to exclude preprints and articles-in-press. The search returned 125 records.

2.3 Inclusion and exclusion criteria

Eligibility criteria were specified a priori and organised into eight inclusion criteria (I1–I8) and eight exclusion criteria (E1–E8), applied across two screening stages.

Criteria I1–I4 and E1–E4 were operationalised automatically through the Scopus filters described in Section 2.2: publication year 2020–2026 (I1/E1), English language (I2/E2), journal article document type (I3/E3), and final publication stage (I4/E4).

Criteria I5–I8 and E5–E8 were applied manually during title/abstract screening and full-text assessment. For inclusion, a study was required to: explicitly address customer or user behaviour within a digital logistics platform or ecosystem context (I5); examine at least one of drivers, barriers, or technology adoption trends relevant to customer mobilization (I6); employ empirical, conceptual, or structured review methodology (I7); and be situated within the logistics, supply chain, or e-commerce industry (I8). Studies were excluded if: customer or user behaviour was not a primary construct of the research (E5); the focus was on conventional or non-digital logistics without a platform or ecosystem element (E6); the article duplicated one already included (E7); or the full text could not be accessed after two retrieval attempts (E8).

Three additional exclusion subcodes were developed inductively during full-text assessment to accommodate patterns not anticipated in the a priori criteria: EX1 (respondents were couriers, managers, or corporate entities rather than end-user customers); EX2 (the study focused on technical systems or infrastructure without examining user behaviour); and EX3 (the digital logistics context was too peripheral to the study’s primary research question to yield meaningful data extraction).

2.4 Study selection process

Study selection proceeded through three sequential stages, documented in the PRISMA 2020 flow diagram ( Figure 1).

Figure 1. PRISMA 2020 flow diagram illustrating the study selection process (adapted from Page et al., 2021).

Stage 1 — Identification. The Scopus search returned 125 records after automated filter application. No duplicate removal was required as the search was conducted within a single database.

Stage 2 — Title and abstract screening. All 125 records were screened against criteria I5–I8 and E5–E8 based on titles, abstracts, and author keywords. This stage yielded 83 records for full-text assessment and 42 exclusions. The primary exclusion reason was E5, applied to studies centred on routing algorithm optimisation, mathematical delivery models, or technical logistics systems without a behavioural component. E6 was applied to studies situated in out-of-scope industry contexts including automotive supply chain, tourism logistics, and LPG distribution.

Stage 3 — Full-text assessment. The 83 remaining records were assessed against the full eligibility criteria. Seventeen articles were excluded (exclusion rate: 20.5%), yielding a final corpus of 66 articles. The most frequent exclusion reason was E5 + EX1, where respondents were delivery couriers, logistics managers, or corporate entities rather than end-user customers (n = 8). Full exclusion reason detail is provided in Appendix D.

2.5 Quality assessment

All 66 articles were assessed using an adapted version of the JBI Critical Appraisal Checklist (Joanna Briggs Institute, 2020). JBI was selected because it provides purpose-built checklists for multiple study designs including cross-sectional surveys, quasi-experimental studies, mixed-method studies, and systematic reviews accommodating the methodological heterogeneity of the included corpus. This makes JBI more appropriate for management and social science SLRs than alternatives designed primarily for clinical research (CASP, ROBIS) or restricted to a single design type (AXIS).

The checklist was adapted to eight items: (Q1) research question clearly defined; (Q2) study design appropriate to objectives; (Q3) population and sample clearly defined and representative; (Q4) measurement instruments valid and reliable; (Q5) data analysis adequate and appropriate; (Q6) findings address the research questions; (Q7) limitations identified and discussed; (Q8) contribution to one or more RQs of this review. Each item was scored dichotomously (Yes = 1, No/Unclear = 0), yielding a maximum score of 8. Quality thresholds were set at High (≥6), Medium (4–5), and Low (≤3).

All 66 articles received a High rating (mean score: 7.71/8; score distribution: 8/8 n = 52, 7/8 n = 9, 6/8 n = 5). No articles were excluded from the synthesis on quality grounds. Articles scoring 6/8 presented minor weaknesses in sample representativeness reporting (Q3), instrument validity documentation (Q4), or discussion of limitations (Q7), but remained eligible given their direct and documented contribution to the RQs (Q8). Full quality assessment scores are reported in Appendix E.

2.6 Data extraction

Data were extracted from all 66 included articles using a standardised extraction form comprising 15 fields: article number, author(s), publication year, title, research method, sample size, theoretical framework, key findings, limitations, and four RQ-specific relevance fields (RQ1: emerging trends; RQ2: key drivers; RQ3: barriers and challenges; RQ4: future directions), supplemented by country/geographic context and primary technology focus. All 66 articles yielded complete data across all 15 fields. The full extraction table is provided in Appendix F.

2.7 Synthesis method

Narrative synthesis was selected as the primary synthesis method, consistent with Popay et al.'s (2006) guidance for heterogeneous evidence bases where methodological diversity precludes statistical pooling. Three reasons precluded meta-analysis: outcome variables were operationalised heterogeneously across studies (adoption intention, customer satisfaction, continuous-use intention, loyalty, and willingness to pay were used as dependent variables but are not statistically equivalent); the corpus employed 18 distinct theoretical frameworks, making effect size comparisons theoretically inappropriate; and only a minority of studies reported standardised effect sizes in a form suitable for pooling.

Narrative synthesis was complemented by two structured analytical procedures. Thematic analysis was applied to RQ2 (drivers) and RQ3 (barriers): driver and barrier constructs were extracted from each article’s RQ-specific field, coded inductively at the construct level, and clustered deductively into theoretically coherent categories using established frameworks from the technology acceptance and service quality literatures as organising schemas. Frequency-based construct mapping was applied across all four RQs to identify convergent evidence (constructs appearing in five or more studies) and divergent evidence (constructs producing contradictory findings across studies). Vote counting was used exclusively to establish construct prevalence and was not used to draw directional causal inferences (Borenstein et al., 2009).

3.1 Overview and characteristics of included studies

The 66 included studies were published between 2020 and 2026. Publication volume increased substantially from 2023 onward, with 35 studies (53.0%) appearing in the 2024–2026 period ( Figure 2). The corpus comprises 53 quantitative studies (80.3%), eight mixed-method studies (12.1%), and five qualitative or expert-based studies (7.6%). The Technology Acceptance Model (TAM) or its extensions is the dominant theoretical framework, employed in 18 studies (27.3%), followed by Service Quality Theory (n = 7, 10.6%), SERVQUAL (n = 4, 6.1%), and UTAUT2 (n = 3, 4.5%) ( Figure 3). Geographically, China contributes the largest single-country share (n = 13, 19.7%), followed by the United States (n = 7, 10.6%), Germany (n = 5, 7.6%), and Singapore (n = 5, 7.6%); the remaining 36 studies span 17 additional countries, representing 21 distinct countries in total. The methodological and contextual characteristics of the included studies are summarised in Table 1.

Figure 2. Distribution of included studies by publication year (n = 66).

Figure 3. Theoretical frameworks employed in included studies (n = 66; studies may employ more than one framework).

3.2 RQ1: Emerging trends in digital logistics customer mobilization (2020–2026)

Thematic analysis identifies five trend clusters across the 66 included studies.

3.2.1 Trend 1: Proliferation of autonomous and contactless delivery technologies

Autonomous delivery robots (ADR), autonomous delivery vehicles (ADV), and delivery drones collectively account for 22 studies (33.3%). Yuen et al. (2022) document consumer acceptance of ADR in Singapore through an integrated TAM-health belief model framework, identifying perceived value as the strongest acceptance determinant. Kaiser et al. (2024) find that performance expectancy and effort expectancy influence ADR acceptance in Germany, with acceptance higher for package delivery than for meal delivery scenarios. Kapser and Abdelrahman (2020) and Kapser et al. (2021) apply extended UTAUT2 models to ADV acceptance in Germany across two studies, identifying price sensitivity as the strongest predictor in 2020 and trust in technology as a primary determinant in 2021. Meyer-Waarden et al. (2026) investigate AI-automated drone delivery in France using a 2 × 2 × 2 experimental design (n = 3,212), finding that environmental concerns enhance the positive effect of performance expectancy on well-being and that product criticality weakens the well-being-adoption intention relationship.

3.2.2 Trend 2: Expansion of platform-based and crowdsourced logistics models

Ta et al. (2023) analyse customer reviews from multiple US retailers using propensity score matching and cognitive appraisal theory, finding that crowdsourced delivery generates higher customer appraisals of timeliness, price, and reliability than traditional delivery. Liu, Y. et al. (2024) apply an integrated e-CSI and PAM-ISC model to O2O crowdsourcing logistics in China (n = 292), finding that trust contributes most to continuous-use intention among all constructs examined. Koh et al. (2024) integrate the health belief model and TAM in a study of 500 Singaporean consumers, finding perceived usefulness as the primary driver of crowdsourced logistics platform adoption. Zhao et al. (2024) introduce a second-order Q-ETAM model to examine errand delivery a novel platform-based delivery form finding that service quality tangibles have the greatest effect on consumer usage intention. Bossong et al. (2025) apply the TOE framework to digital intermodal freight platforms through multiple case studies, identifying nine levers for platform adoption and documenting opposing preferences between sellers and demanders of freight services. Permatasari et al. (2025) evaluate the SILCARE digital logistics platform for MSMEs in Indonesia using SUS assessments and DBSCAN clustering (n = 100), identifying four distinct user behaviour clusters with 37% of users struggling with complex workflows.

3.2.3 Trend 3: Sustainability and green logistics as mainstream consumer concerns

Klein and Popp (2022) test a TAM-sustainability model with 536 German online buyers, finding that all three sustainability dimensions environmental, economic, and social significantly influence acceptance of home delivery, parcel lockers, and click-and-collect, while perceived costs constrain acceptance across all three methods. Rizzitello et al. (2026) conduct a controlled experiment partnered with an Italian supermarket, finding that moral green nudges outperform financial incentives in driving sustainable delivery choices and that peer influence amplifies adoption of shared delivery systems. Yang et al. (2023) apply contingency, transaction cost, and trust theories to 453 users of green shopping platforms in South Korea, finding that green cooperation positively affects consumer trust and sustainable consumption behaviour. Raźniewska et al. (2026) develop and test an integrated Technology Readiness-TAM model, finding that perceived ease of use significantly impacts perceived usefulness and that environmental awareness improves perceptions of EV courier services without directly translating into adoption intent. Kiba-Janiak et al. (2022) apply conjoint analysis to Polish e-customers, finding that delivery speed and low cost dominate preferences over sustainability options and that most customers select home delivery despite the availability of sustainable alternatives.

3.2.4 Trend 4: Service quality re-conceptualisation toward experience and trust

Vrhovac et al. (2023) develop and validate the CMX-LMD scale (n = 907) through EFA and CFA, identifying six customer experience dimensions: delivery efficiency, parcel tracking, smooth delivery, visual appeal, joyful anticipation, and convenience. Vrhovac et al. (2024) apply SEM to 907 participants across three questionnaires, identifying trust in courier service, delivery speed, price, and courier call before delivery as significant satisfaction predictors, with visual appeal as a negative predictor. Masorgo et al. (2023) conduct a scenario-based vignette experiment finding that inappropriate driver behaviour activates anger and driver inflexibility activates sadness, with both emotional responses reducing customer repurchase intentions; the inflexibility effect is weaker for outsourced than for private fleet drivers. Uvet et al. (2024) develop and test a hybrid LSQ model incorporating a second-order assurance quality construct comprising personnel contact quality, order discrepancy handling, and returns management finding that online ordering procedures and assurance quality account for more variance in customer satisfaction than conventional LSQ dimensions.

3.2.5 Trend 5: Growing geographic diversification of research

Beyond the dominant China-US-Germany-Singapore cluster, the corpus includes studies from Kenya (Mogire et al., 2023), South Africa (Heyns & Kilbourn, 2022), Indonesia (Monoarfa et al., 2024; Permatasari et al., 2025), Vietnam (Do et al., 2023; Trung et al., 2025), Ethiopia (Tadesse et al., 2025), Saudi Arabia (Alahmari et al., 2025), Egypt (Sultan et al., 2023), Iran (Ganjipour & Edrisi, 2023; Esmaili et al., 2025), Poland (Kiba-Janiak et al., 2022), Sweden (Patowary et al., 2023), South Korea (Yang et al., 2023; Jang et al., 2024), Italy (Iannaccone et al., 2021; Rizzitello et al., 2026), France (Meyer-Waarden et al., 2026), India (Sharma et al., 2022; Pathak et al., 2023), Turkey (Toraman & Oz, 2023; Toraman et al., 2024), Malaysia (Teng et al., 2024; Aziz et al., 2025), and Taiwan (Bossong et al., 2025). Tadesse et al. (2025) represent the first application of TAM to digital logistics in a low-income country context, finding that infrastructure availability, human resource capacity, technological accessibility, and supportive policies significantly influence adoption intention beyond the core TAM constructs.

3.3. RQ2: Key Drivers influencing customer mobilization in digital logistics ecosystems

Thematic analysis identifies 47 distinct driver constructs across the 66 studies, consolidated into five clusters. The frequency distribution of the ten most consistently cited driver constructs is presented in Table 2.

3.3.1 Driver cluster 1: Technology acceptance constructs (TAM Core)

Perceived usefulness is the most consistently documented driver, appearing in 17 studies across drone delivery (Leon et al., 2025; Koh et al., 2023; Toraman & Oz, 2023), ADR and ADV adoption (Yuen et al., 2022; Esmaili et al., 2025; Zhang et al., 2025), crowdsourced logistics (Koh et al., 2024), and digital logistics platform adoption (Liu et al., 2025). Perceived ease of use is documented in 10 studies, with Raźniewska et al. (2026) confirming a significant sequential pathway from perceived ease of use to perceived usefulness. Several studies document perceived usefulness as a mediating construct: environmental and health dispositions enhance perceived usefulness (Leon et al., 2025); health concerns shape usefulness perceptions in crowdsourced logistics (Koh et al., 2024); flow experience mediates the path from DOI constructs to platform adoption through usefulness (Liu et al., 2025); and service quality perceptions shape usefulness evaluations in errand delivery platforms (Zhao et al., 2024).

3.3.2 Driver cluster 2: Service quality and experiential dimensions

Service quality dimensions are documented in 10 studies across multiple geographic contexts. Trung et al. (2025) find that delivery staff interaction has the greatest single impact on perceived value in the Mekong Delta (n = 414). Do et al. (2023) identify delivery time as the most critical satisfaction determinant for Generation Z consumers in Vietnam (n = 510). Wang et al. (2024) identify reliability, convenience, freshness, and personnel contact quality as the four key LSQ dimensions in fresh product e-commerce (n = 222). Monoarfa et al. (2024) demonstrate that E-LSQ mediates between user confirmation and customer satisfaction in fresh produce e-commerce in Indonesia (n = 364). Tandon and Ertz (2024) confirm E-LSQ as a stronger antecedent of customer satisfaction than gamification in online shopping (n = 634). Aziz et al. (2025) find empathy and assurance significant but reliability and responsiveness non-significant in the Malaysian courier context (n = 245).

3.3.3 Driver cluster 3: Trust, perceived value, and relational drivers

Trust is a significant driver in seven studies. Liu, Y. et al. (2024) find that trust contributes most to continuous-use intention and significantly reduces perceived risk in O2O crowdsourcing logistics (n = 292). Kapser et al. (2021) identify trust in technology as a primary ADV acceptance determinant (n = 501). Yang et al. (2023) document that green cooperation positively affects consumer trust, which in turn drives sustainable consumption behaviour and loyalty (n = 453). Jarutirasarn and Thirapatsakun (2025) document a satisfaction-sufficiency phenomenon in which positive customer experience diminishes the engagement-loyalty link in Thailand’s e-commerce logistics sector (n = 500). Perceived value operates as a mediating driver in Trung et al. (2025), where delivery quality dimensions drive perceived value which drives satisfaction and loyalty, and in Liu, Y. et al. (2024), where perceived value mediates service quality and continuous-use intention.

3.3.4 Driver cluster 4: Social, normative, and behavioural drivers

Attitude is documented as a significant mediator between beliefs and behavioural intention in Wang et al. (2020), where three competing models demonstrate that no attitude-free specification achieves equivalent fit. Yuen et al. (2022) find attitude holds the largest total effect on ADR adoption intention in Singapore (n = 500). Liu, S. et al. (2024) confirm that attitude directly and indirectly mediates SST adoption intention in China. Social influence is significant in Kapser and Abdelrahman (2020, 2021) and Zhong et al. (2022). Subjective norms are confirmed significant in Esmaili et al. (2025) (n = 1,567). Ganjipour and Edrisi (2023) find that problem awareness, ascribed responsibility, and environmental concern positively influence personal norm, which in turn affects intention to use delivery robots (n = 463). Relative advantage and compatibility from Diffusion of Innovation theory are confirmed as drivers in Wang et al. (2020), Zhang et al. (2025), Liu, S. et al. (2024), and Ganjipour and Edrisi (2023).

3.3.5 Driver cluster 5: Contextual and hedonic drivers

Hedonic motivation is significant in ADV adoption (Kapser & Abdelrahman, 2020; Kapser et al., 2021) and EV delivery acceptance by small businesses (Toraman et al., 2024). Liu et al. (2025) find that flow experience demonstrates a stronger mediating effect on digital logistics platform adoption than technology perception alone in the maritime sector. Encarnación et al. (2025) find that workplace proximity is a stronger parcel locker adoption predictor than home proximity. Iannaccone et al. (2021) quantify willingness to pay for locker accessibility in Rome, finding distance and accessibility as the main choice determinants among young consumers. Personal innovativeness is confirmed as a significant moderator of APL adoption by Jang et al. (2024) (n = 459).

3.4 RQ3: Barriers and challenges hindering effective customer mobilization

Thematic analysis identifies 41 distinct barrier constructs, consolidated into four clusters. The frequency distribution of the ten most consistently cited barrier constructs is presented in Table 3.

3.4.1 Barrier cluster 1: Technological and perceptual barriers

Complexity is the most consistently documented technological barrier, appearing in seven studies across digital logistics platforms (Liu et al., 2025; Sultan et al., 2023), ADV and ADR adoption (Zhang et al., 2025; Ganjipour & Edrisi, 2023), self-service technology (Liu, S. et al., 2024), and last-mile logistics applications (Sultan et al., 2023). Sultan et al. (2023) identify complexity, insufficient user-developer collaboration, and limited technical knowledge as the three dominant barriers to logistics application adoption in Egypt (n = 1,060). Privacy concerns are documented in five studies, with Garg et al. (2025) finding that women exhibit heightened privacy sensitivity in healthcare drone delivery contexts through PLS-SEM and fsQCA analysis. Esmaili et al. (2025) document perceived theft risk as a novel barrier in ADV adoption in Iran (n = 1,567), with the majority of qualitative respondents citing theft risk as the primary reason they would not use ADVs in urban settings. Low perceived usefulness functions as a barrier in contexts of limited consumer awareness of new delivery modalities (Tadesse et al., 2025; Jang et al., 2024).

3.4.2 Barrier cluster 2: Economic and structural barriers

Klein and Popp (2022) find that perceived costs constrain consumer acceptance across all three delivery methods examined (n = 536). Kapser and Abdelrahman (2020) identify price sensitivity as the strongest single predictor of ADV acceptance in their UTAUT2 model (n = 501). Amaya et al. (2026) find that cost sensitivity is the strongest deterrent to continued online shopping under retail delivery fee conditions, with perceived fairness emerging as a necessary condition for maintaining shopping behaviour. Tadesse et al. (2025) find that infrastructure availability, human resource capacity, and technological accessibility significantly moderate TAM core constructs in Ethiopia. Sharma et al. (2022) identify I4.0 infrastructure as the most fundamental prerequisite for smart last-mile delivery in India, with delivery capacity as the primary operational barrier. Pathak et al. (2023) apply DEMATEL analysis to 20 barriers, identifying routing simulation inadequacy and lack of alternative vehicles as the two cause-group barriers with the greatest downstream impact. Mogire et al. (2023) and Trung et al. (2025) both document returns management as the dimension with the lowest customer satisfaction scores across their respective samples in Kenya and Vietnam.

3.4.3 Barrier cluster 3: Socio-behavioural barriers

Kiba-Janiak et al. (2022) find through conjoint analysis that most Polish e-customers select home delivery despite the availability of sustainable alternatives, attributing this pattern to low ecological awareness. Patowary et al. (2023) document a significant gap between available delivery options and consumer preferences in Sweden, with home delivery preferred despite expressed sustainability concerns. Rizzitello et al. (2026) demonstrate experimentally that financial incentives alone are insufficient to overcome home delivery inertia. Jang et al. (2024) establish distrust as a significant inhibitor of APL adoption operating through its negative effects on perceived usefulness and ease of use (n = 459). Liu, Y. et al. (2024) confirm that perceived risk negatively moderates the satisfaction-continuance intention relationship in crowdsourcing logistics (n = 292). Masorgo et al. (2023) find through scenario-based experiment that inappropriate driver behaviour produces anger and driver inflexibility produces sadness, both reducing repurchase intentions. Zhang et al. (2025) document social awkwardness as a novel barrier specific to ADV adoption in Chinese contexts, where norms around human-robot interaction create adoption resistance not captured by conventional TAM constructs.

3.4.4 Barrier cluster 4: Institutional and regulatory barriers

Tadesse et al. (2025) find that supportive government policies significantly moderate TAM core constructs in Ethiopia, establishing that digital logistics adoption in low-income contexts cannot be adequately explained by individual-level TAM alone. Mutambik (2025) identifies regulatory frameworks as one of four thematic areas critical to the trajectory of last-mile delivery by 2030, drawing on a two-round Delphi study with 52 logistics experts. Bossong et al. (2025) document market transparency fears among sellers of intermodal freight services as a platform-governance barrier through multiple case studies, finding that sellers fear market transparency while demanders favour it a structural conflict not resolved by existing platform design solutions.

3.5. RQ4: Research gaps and future directions

Analysis of author-stated future directions across the 66 studies and cross-study gap identification yields five systematic research gap clusters, summarised in Table 4.

3.5.1 Gap 1: Longitudinal evidence and post-adoption dynamics

Thirteen studies explicitly call for longitudinal investigation. Cross-sectional designs account for 83% of the quantitative studies in the corpus. Specific longitudinal gaps include post-pandemic persistence of contactless delivery adoption (Koh et al., 2024; Jang et al., 2024); longitudinal tracking of the sustainability attitude-behaviour gap (Klein & Popp, 2022; Raźniewska et al., 2026); trust formation and erosion dynamics in crowdsourcing logistics (Liu, Y. et al., 2024); and multi-timepoint validation of the CMX-LMD scale (Vrhovac et al., 2023).

3.5.2 Gap 2: Cross-cultural and multi-country comparative studies

Nine studies call for cross-cultural validation. TAM universality is questioned by Tadesse et al. (2025), who find external factors beyond perceived usefulness and ease of use critical in Ethiopia. Cross-cultural comparison of ADV and drone adoption is absent — existing studies are limited to Germany, Singapore, China, and Iran, with no multi-country comparison. Aziz et al. (2025) find different service quality dimensions significant in Malaysia than in European and US contexts. Parcel locker adoption outside Asian and European markets remains unstudied.

3.5.3 Gap 3: Policy, regulatory, and institutional research

Seven studies identify policy and institutional research as a priority. Regulatory frameworks for drone and ADV deployment are absent despite extensive consumer perception data (Leon et al., 2025; Meyer-Waarden et al., 2026). Policy instrument design for sustainable last-mile logistics is identified by Rizzitello et al. (2026) and Kiba-Janiak et al. (2022), who note the absence of scaled policy interventions. Logistics digitalization policy in low-income countries is identified by Tadesse et al. (2025) as a critical gap. Data governance and privacy regulation in platform-based logistics remains unaddressed despite privacy concerns documented in five studies.

3.5.4 Gap 4: Supply-side and ecosystem-level analysis

Bossong et al. (2025) provide the only study in the corpus addressing supply-side adoption dynamics in digital freight platforms. No study simultaneously models multi-actor digital logistics ecosystem dynamics. Network effects and critical mass dynamics in digital logistics platforms have not been empirically investigated. Carrier and courier adoption dynamics beyond electric vehicle acceptance remain largely unexamined.

3.5.5. Gap 5: Emerging technology and methodological gaps

Technology gaps include blockchain applications in logistics consumer trust, AI-driven personalisation in logistics platforms (Alahmari et al., 2025), and gamification design — where Tandon and Ertz (2024) find no effect on satisfaction while Liu, X. et al. (2024) document significant gamification affordance effects, indicating design-contingent effectiveness. Methodological gaps include machine learning analysis of consumer behaviour (one study: Aljohani, 2024), configurational fsQCA approaches (one study: Garg et al., 2025), and big data analytics integration (one study: Shi et al., 2021, n = 54,250 reviews).

4.1 Integrative interpretation of findings

The synthesis of 66 peer-reviewed studies across four research questions reveals four cross-cutting patterns that extend beyond the RQ-specific findings and warrant theoretical interpretation.

Perceived usefulness as universal mediating hub. The most robust finding of this review is that perceived usefulness functions not as a terminal attitudinal construct but as a mediating hub through which multiple antecedent variables environmental dispositions, health beliefs, flow experience, service quality perceptions, and social norms exert their influence on adoption intention. This hub-and-spoke architecture is documented consistently across all technology types and across 21 countries, suggesting that perceived usefulness occupies a structurally privileged position in the causal architecture of digital logistics customer mobilization that goes beyond its original conceptualisation in Davis's (1989) TAM. This finding extends Venkatesh et al.'s (2003) UTAUT argument that perceived usefulness is the most powerful predictor of behavioural intention, by demonstrating that in digital logistics contexts, usefulness also absorbs and mediates upstream effects from non-TAM constructs including environmental concern (Leon et al., 2025), health risk perception (Koh et al., 2024), and hedonic flow experience (Liu et al., 2025). The implication for platform intervention design is direct: strategies that enhance usefulness perception through value demonstration, onboarding quality, and risk reduction are more generalisable and more evidence-based than strategies targeting any single upstream construct in isolation.

The trust-complexity nexus as compound barrier dynamic. Trust deficits and complexity emerge as the two dominant barrier types in the corpus, but their co-occurrence is not coincidental. Both operate through distinct psychological mechanisms trust deficits through relational and emotional pathways, complexity through cognitive pathways yet both ultimately suppress adoption intention. Several studies document their structural interaction: Sultan et al. (2023) find that insufficient user-developer collaboration simultaneously creates complexity barriers through poor interface design and trust barriers through perceived developer indifference; Jang et al. (2024) find that distrust in APL systems operates specifically through its negative effect on perceived ease of use, creating a cognitive-relational compound in which emotional distrust manifests as cognitive difficulty. This trust-complexity nexus has not been theorised in the prior TAM or trust literature, where both constructs are typically modelled as independent predictors. The present synthesis suggests that barrier mitigation strategies addressing only one construct are likely to be suboptimal, and that effective customer mobilization requires simultaneous intervention on both relational and cognitive dimensions a finding that challenges the common industry practice of UX simplification as a standalone adoption strategy.

The sustainability aspiration-adoption gap. Klein and Popp (2022), Raźniewska et al. (2026), and Kiba-Janiak et al. (2022) all document that environmental awareness improves consumers’ perceptions of sustainable delivery options without directly driving adoption intent, and that home delivery inertia persists even when consumers articulate sustainability values and sustainable alternatives are readily available. This pattern replicates the attitude-behaviour gap documented in the broader consumer behaviour literature (Kollmuss & Agyeman, 2002) and extends it specifically to the digital logistics domain. Critically, Rizzitello et al. (2026) provide experimental evidence that social pressure in the form of moral green nudges and peer influence rather than financial incentives is required to close this gap, with their shared cart mechanism producing cost reductions of 76.8% when social normative mechanisms are activated. This finding has direct implications for how logistics platform sustainability strategies should be designed: economic incentive mechanisms alone are insufficient, and social normative architecture peer visibility, collective delivery options, and moral framing is the appropriate intervention lever.

Methodological dominance and confirmation risk. The dominance of TAM-based, cross-sectional, single-country, quantitative survey studies 53 of 66 studies (80.3%) creates a risk of systematic confirmation bias. TAM constructs will tend to be found significant when TAM is the primary lens, cross-sectional designs cannot establish temporal precedence, and single-country samples prevent assessment of cultural boundary conditions. The isolated but theoretically significant contributions of studies using Privacy Calculus Theory (Leon et al., 2025; Koh et al., 2023), Health Belief Model (Yuen et al., 2022; Koh et al., 2024), Cognitive Appraisal Theory (Masorgo et al., 2023), and configurational fsQCA (Garg et al., 2025) demonstrate that alternative theoretical lenses reveal dimensions of customer mobilization that TAM-centric research systematically underestimates particularly emotional, normative, and configurational pathways to adoption.

4.2 Theoretical contributions

This review makes five theoretical contributions to the digital logistics and customer mobilization literature.

First, it provides the first comprehensive cross-technology, cross-geography synthesis of customer mobilization in digital logistics ecosystems. Prior reviews have been limited to single technology types drone delivery (Zrelli et al., 2024), autonomous vehicles, or parcel lockers single regions, or narrow constructs. By synthesising evidence across all digital logistics technology categories and 21 countries, this review establishes a unified evidence base that identifies which findings are robust across contexts and which are context-contingent.

Second, it advances TAM theory in the digital logistics domain by recharacterising perceived usefulness from a terminal construct to a mediating hub with a complex upstream causal architecture. This recharacterisation has implications for how TAM-based research should be designed and interpreted: studies that model perceived usefulness only as an independent or mediating variable without examining what shapes it are likely to underestimate its theoretical richness and practical leverage.

Third, it introduces the trust-complexity nexus as a theoretically novel compound barrier dynamic that bridges the technology acceptance literature and the trust literature. By demonstrating that trust deficits and complexity interact rather than operate independently, this review offers a more complete explanation of adoption resistance in digital logistics contexts than either construct provides alone, and points toward integrated mitigation strategies that current practice has not yet operationalised.

Fourth, it extends the attitude-behaviour gap theory (Kollmuss & Agyeman, 2002) to the digital logistics domain and identifies social-normative mechanisms rather than economic incentive mechanisms as the primary lever for closing the sustainability aspiration-adoption gap. This extends existing theoretical work on pro-environmental behaviour by demonstrating that the gap is domain-specific and that the mechanisms required to close it differ from those effective in other consumption domains.

Fifth, it documents the limits of TAM universality by establishing, through the Ethiopian case (Tadesse et al., 2025), that external institutional factors infrastructure availability, human resource capacity, and supportive policies significantly moderate core TAM constructs in low-income country contexts. This contributes to the ongoing debate about the cultural and institutional boundary conditions of TAM (Bagozzi, 2007) and suggests that technology acceptance frameworks developed in high-income market contexts require contextual adaptation before application in developing economies.

4.3 Managerial and practical implications

The findings carry actionable implications for four practitioner groups.

For digital logistics platform operators, the centrality of perceived usefulness as a mediating hub implies that platform strategy should prioritise value communication and demonstration over feature complexity. Onboarding experiences that rapidly demonstrate tangible utility through personalised use cases, visible performance data, and comparative advantage framing are more likely to accelerate customer mobilization than interface aesthetics or feature richness. The trust-complexity nexus further implies that UX simplification and trust-building must be pursued simultaneously: platforms that reduce cognitive load without addressing trust concerns, or vice versa, will encounter persistent adoption resistance at the barriers they have not addressed.

For last-mile delivery service providers, the reconceptualisation of service quality toward experiential and relational dimensions has direct operational implications. Trust in courier service surpasses delivery speed and price as the dominant satisfaction driver (Vrhovac et al., 2024), implying that investment in driver conduct, communication quality, and empathy training yields higher customer mobilization returns than further speed optimisation in markets where baseline delivery time performance is already acceptable. Returns management is consistently identified as the weakest service quality dimension across multiple geographic contexts, representing a high-leverage improvement opportunity. For providers deploying parcel lockers, workplace proximity should be prioritised over residential proximity in location strategy.

For logistics platform policymakers, three gaps require urgent attention. Regulatory frameworks for drone delivery and ADV deployment lag significantly behind consumer acceptance levels, creating a policy vacuum that constrains commercial deployment. Green nudge policy instruments particularly peer influence mechanisms and moral framing are more effective than delivery fee mechanisms for shifting consumer delivery behaviour toward sustainable alternatives, with experimental evidence from Rizzitello et al. (2026) providing a direct model for policy design. In low-income country contexts, logistics digitalization policy must address infrastructure and capacity constraints as prerequisites rather than treating technology adoption as an individual-level behavioural challenge.

For platform investors and startups, particularly in emerging markets such as Saudi Arabia, Indonesia, and Vietnam, service performance and digital integration are the dominant customer experience predictors (Alahmari et al., 2025; Monoarfa et al., 2024; Trung et al., 2025), suggesting that operational consistency should precede marketing investment. MSME-specific platform design must prioritise usability over feature richness, given that 37% of MSME users in Permatasari et al. (2025) struggled with complex workflows even after usability improvements.

4.4 Limitations

Several limitations of this review should be considered when interpreting its findings.

The search was conducted in a single database (Scopus). While Scopus provides comprehensive coverage of peer-reviewed logistics and management journals and is widely used as the sole database in SLRs in this domain (Durach et al., 2017), studies indexed exclusively in Web of Science, Google Scholar, or non-indexed practitioner publications are not represented in the corpus. This limitation may result in underrepresentation of practitioner-oriented evidence and of research published in non-Scopus-indexed regional journals, particularly from Southeast Asian and African contexts.

The restriction to English-language publications excludes a potentially substantial body of logistics and e-commerce research published in Chinese, German, and other languages. Given that China contributes the largest single-country share of the included corpus (19.7%), Chinese-language publications may contain evidence that complements or qualifies the English-language findings, particularly regarding platform-specific consumer behaviour in domestic Chinese logistics ecosystems.

Full-text screening was conducted based on extended abstracts from Scopus rather than physical full-text access, creating the possibility of residual misclassification in borderline cases. While the 20.5% full-text exclusion rate is within acceptable SLR ranges (Liberati et al., 2009), studies that were included based on abstract content but whose full text would not have met the eligibility criteria cannot be entirely ruled out.

Quality assessment was conducted by a single reviewer with a consensus protocol for borderline cases, rather than by two independent reviewers with inter-rater reliability calculation. While all 66 included articles received High ratings on the JBI checklist, the absence of independent verification means that systematic reviewer bias in the quality scoring cannot be excluded.

Finally, narrative synthesis does not permit quantitative effect size comparison. Where directional findings are reported for example, that perceived usefulness is the most frequently cited driver these represent construct prevalence across studies rather than pooled effect sizes, and should not be interpreted as evidence of comparative effect magnitude.