AI helps small firms but rarely does the heavy lifting alone: a review of 29 studies shows performance gains depend on data governance, complementary capabilities and financing, with productivity and employment effects varying sharply by firm type and context.

ARTIFICIAL INTELLIGENCE ADOPTION AND FIRM PERFORMANCE IN SMES A SYSTEMATIC LITERATURE REVIEW AND FUTURE RESEARCH AGENDA

Ridha Rayan Furqan, Wiwiek Rabiatul Adawiyah, Ali Şahin, Ratno Purnomo · August 10, 2026 · The International Conference on Sustainable Economics Management and Accounting Proceeding

openalex review_meta medium evidence 7/10 relevance Full text usable extracted full text DOI Source PDF

Structured author observations

Linked only from stored provider relations; the raw author line above is never matched by name.

OpenAlex

Latest observation: August 24, 2026

Ridha Rayan Furqan provider ID
Wiwiek Rabiatul Adawiyah provider ID
Ali Şahin provider ID
Ratno Purnomo exact ORCID

Semantic Scholar

Latest observation: August 24, 2026

Ridha Rayan Furqan provider ID
W. Adawiyah provider ID
Ali Şahin provider ID
Ratno Purnomo provider ID

A systematic review of 29 studies finds AI adoption improves SME performance only when paired with complementary capabilities — knowledge management, human capital, governance and financing — and that effects are heterogeneous across firm size, industry and institutional context.

Citation observations

Cumulative provider counts captured on specific dates; providers are never combined.

0 cumulative citations

OpenAlex · Observed August 24, 2026

View corpus context

0 cumulative citations

Semantic Scholar · Observed August 24, 2026

View corpus context

The relationship between artificial intelligence (AI) adoption and firm performance in small and medium-sized enterprises(SMEs) has attracted growing scholarly attention, yet the literature remains theoretically fragmented and dominated by techno-optimistic assumptions that portray AI as a direct pathway to superior performance. This study addresses two central research questions: why do AI adoption outcomes remain heterogeneous across SMEs, and through what organizational mechanisms does AI create business value? A systematic search of the Scopus database, followed by PRISMA-guided screening, yielded 29peer-reviewed articles analyzed through VOS viewer bibliometric mapping and qualitative content analysis. Four thematic clusters were identified: AI Adoption and SME Sustainable Performance; AI Investment and Firm-Level Heterogeneity; Green Intellectual Capital and Innovation; and AI Application and Service Productivity in China. Drawing on these clusters, an integrative Antecedents-Mediators-Moderators-Outcomes (AMMO) framework is proposed. Findings demonstrate that AI adoption outcomes are contingent upon complementary organizational capabilities, knowledge management infrastructure, human capital quality, and institutional context rather than technology deployment alone. Five value-creating mechanism categories are identified: innovation and transformation, human capital and knowledge, operational and productivity, strategic and behavioral, and financial and investment mechanisms. The study contributes by challenging techno-deterministic assumptions, advancing the AMMO framework as a conceptual architecture for future inquiry, and positioning green intellectual capital as an emerging theoretical frontier. Implications for managers, policymakers, and future researchers are discussed.

Summary

Main Finding

AI adoption in SMEs does not automatically translate into superior firm performance. Outcomes are heterogeneous and depend on complementary organizational capabilities (knowledge management, human capital, data governance), firm- and context-level conditions (size, ownership, industry, financing constraints, institutional environment), and process-level mechanisms. The authors synthesize the literature into an Antecedents–Mediators–Moderators–Outcomes (AMMO) framework and identify five classes of value-creation mechanisms through which AI can affect SME performance.

Key Points

Literature scope and size
- Systematic review of 29 peer‑reviewed Scopus articles (through 19 May 2026) using PRISMA screening.
- Research concentrated in 2024–2026 (≈90% of studies), indicating a nascent but rapidly growing field.
- Cross‑disciplinary outlets; Sustainability and sustainability‑oriented journals are prominent; most studies are in Q1–Q2 journals.
Thematic clusters (from bibliometric mapping)
AI Adoption & SME Sustainable Performance: PLS‑SEM and RBV/dynamic capabilities framing; emphasis on mediating roles of knowledge, intellectual capital, absorptive capacity.
AI Investment & Firm‑Level Heterogeneity: panel/difference‑in‑dif studies showing heterogeneous effects (skill‑biased adjustments, financing constraints matter).
Green Intellectual Capital & Innovation: GIC mediates AI → green innovation → sustainable performance; AI more an amplifier of knowledge assets than direct innovator.
AI Application & Service Productivity in China: panel and text‑analysis studies finding productivity/R&D gains, labor‑structure effects, and heterogeneity by firm lifecycle and digital maturity.
AMMO framework
- Organizes factors as Antecedents (e.g., AI investment, strategy, digital finance), Mediators (knowledge capability, green intellectual capital, supply‑chain resilience), Moderators (institutional context, financing constraints, digital readiness), and Outcomes (operational, innovation, sustainability, financial performance).
Value‑creating mechanism categories
Innovation & transformation (new products, digital business models)
Human capital & knowledge (talent upgrading, tacit knowledge sharing)
Operational & productivity (process automation, response times, TFP gains)
Strategic & behavioral (decision quality, competitive positioning)
Financial & investment (ESG impacts, financing interactions)
Representative empirical findings
- Chatbot deployments in micro‑firms: ~45.9% reduction in response time; ~14.5% increase in customer satisfaction.
- AI increases high‑skill employee shares and can displace low‑skill roles; net labor effects vary (substitution vs creation).
- Productivity: specialized/refined/innovative SMEs show ~6.82% productivity increase per SD in AI application (China studies).
- AI investment can improve ESG/green innovation but may weaken internal governance controls in some contexts.
Critical stance
- Challenges techno‑deterministic views: AI’s payoff is contingent, not automatic.
- Calls attention to data quality, interoperability, algorithmic bias, and trust as practical barriers.

Data & Methods

Review method
- Systematic literature review following PRISMA and an SLR protocol; combined quantitative bibliometrics (VOSviewer co‑word, overlay/density maps) and qualitative content analysis.
- Search restricted to Scopus, English language, peer‑reviewed journals; final sample = 29 articles after multi‑stage filtering.
Dominant empirical approaches in reviewed literature
- PLS‑SEM and cross‑sectional survey analysis (common for adoption/performance mediation studies).
- Panel fixed effects, instrumental variables, and difference‑in‑differences (for causal/firm‑level heterogeneity work).
- Text‑analysis based measures of AI application and firm‑level AI input indices.
- Case studies and qualitative implementation accounts (micro‑enterprise examples).
Methodological gaps noted
- Few longitudinal causal designs; limited cross‑country comparative work; uneven AI measurement (input vs application vs intensity); underuse of experimental or granular administrative data in SMEs.
Limitations of the review (authors’ acknowledgements)
- Scopus + English restriction may omit relevant regional or non‑English work.
- Small, rapidly evolving literature limits generalizability; publication recency may bias towards short‑term effects.

Implications for AI Economics

For researchers
- Move beyond techno‑deterministic “AI → performance” models. Operationalize the AMMO framework empirically: measure antecedents (strategy, finance), mediators (GIC, knowledge flows), and moderators (institutional/digital readiness).
- Prioritize causal identification (panel IV, diff‑in‑diff, field experiments) and longitudinal studies to capture dynamic effects and learning.
- Improve AI measurement: distinguish investment inputs, deployed applications, intensity, and model transparency/quality. Use firm‑level administrative/transactional data where possible.
- Investigate heterogeneity systematically (size, ownership, industry, lifecycle, country institutions). Study distributional labor effects (skill upgrading vs displacement) in SMEs.
- Explore green intellectual capital (GIC) as an emerging theoretical frontier—how AI interacts with knowledge assets to produce sustainability and innovation returns.
- Expand geographic diversity beyond China and the present dominant publishing venues; include low‑ and middle‑income contexts and micro‑enterprise evidence.
For policymakers
- Support complementary capabilities (training, data infrastructure, digital finance) rather than only subsidizing AI purchases. Financing constraints and governance gaps can mute AI benefits, especially for SMEs.
- Promote data quality, interoperability standards, and small‑firm access to shared AI tools/platforms to reduce implementation risk.
- Design labor and social policies anticipating skill‑biased transitions (retraining, incentives for inclusive adoption).
For managers of SMEs
- Treat AI as a capability that requires complementary investments in human capital, data governance, and process integration. Incremental, context‑sensitive deployments (e.g., focused chatbots or forecasting) can yield measurable gains when matched to firm needs.
- Monitor potential governance and bias risks; align AI adoption with sustainability/green knowledge assets to extract broader value.
Broader economic research agenda
- Quantify aggregate vs distributional effects of SME AI adoption on productivity, employment, and inequality.
- Model interactions between digital finance, AI investment, and innovation returns—test for tipping points where complementarities unlock larger productivity gains.
- Incorporate environmental and ESG outcomes into economic evaluations of AI to assess sustainability trade‑offs and co‑benefits.

Summary: This SLR reframes AI in SMEs as a contingent, capability‑mediated phenomenon, not a plug‑and‑play productivity lever. Future AI economics work should emphasize causal identification, heterogeneity, complementary assets (especially green intellectual capital), and the institutional and financial conditions that determine whether AI generates sustainable firm‑level value.

Assessment

Paper Typereview_meta Evidence Strengthmedium — This is a systematic literature review synthesizing 29 peer-reviewed articles; the underlying primary studies are mixed — many are cross-sectional PLS-SEM studies and qualitative cases, with a smaller set using panel methods, fixed effects, difference-in-differences, or IV. The review identifies consistent conditional patterns but cannot itself establish new causal estimates. Methods Rigormedium — The authors follow an explicit SLR protocol (PRISMA-style screening), use Scopus as a comprehensive source, apply bibliometric mapping (VOSviewer), and perform qualitative content analysis; however, the search is limited to Scopus and English-language peer-reviewed journals, the final sample is small (29 articles), and the review relies heavily on reporting of heterogeneous primary study designs rather than formal meta-analytic causal synthesis. SampleA systematic review of 29 peer-reviewed journal articles indexed in Scopus up to 19 May 2026; included studies span qualitative case studies, cross-sectional PLS-SEM analyses, and several panel and quasi-experimental designs (fixed effects, DiD, IV) with geographic coverage including China, Vietnam, Saudi Arabia and other contexts; journals are primarily Q1–Q2 outlets. Themesadoption productivity innovation human_ai_collab labor_markets GeneralizabilitySearch limited to Scopus-indexed, English-language, peer-reviewed journals — excludes gray literature and non-English studies, Small final sample (29 articles) concentrated in 2024–2026 implies recency bias, Overrepresentation of sustainability-oriented journals may skew thematic emphasis, Geographic concentration in China and some emerging markets limits global representativeness, Heterogeneous and often cross-sectional primary methods constrain causal generalizability, Exclusion of conference papers and working papers may omit emerging empirical findings

Claims (10)

Claim	Direction	Outcome	Confidence & Evidence	Details
The systematic review found that AI adoption outcomes in SMEs are contingent on complementary organizational capabilities, knowledge-management infrastructure, human-capital quality, and institutional context rather than technology deployment alone. Firm Productivity	mixed	Firm performance outcomes associated with AI adoption	Reading fidelity high Study strength medium	n=29 0.24
AI adoption enhances SME performance through capability-mediated pathways in which knowledge management, intellectual capital, and competitive advantage act as sequential mediators. Firm Productivity	positive	Sustainable organizational performance	Reading fidelity high Study strength medium	n=29 0.24
AI usage intensity alone does not guarantee performance improvement; complementary dynamic capabilities, data governance, and human capital are necessary for realizing performance gains. Firm Productivity	mixed	Firm performance improvement following AI use	Reading fidelity high Study strength medium	n=29 0.24
Corporate AI investment increases the proportion of technical employees while reducing non-technical roles, with financing constraints amplifying negative employment effects, particularly among SMEs. Employment	negative	Employment composition and employment levels	Reading fidelity high Study strength medium	n=29 0.24
AI application has no statistically significant aggregate effect on labor demand among Chinese small and micro enterprises because labor-substitution and labor-creation effects offset each other. Employment	null_result	Aggregate labor demand	Reading fidelity high Study strength medium	n=29 statistically insignificant aggregate effect 0.24
Context-sensitive AI chatbot implementation in micro-enterprises can reduce response time by 45.9% and increase customer satisfaction by 14.5%. Organizational Efficiency	positive	Customer-response time and customer satisfaction	Reading fidelity high Study strength low	45.9% reduction in response time; 14.5% increase in customer satisfaction 0.12
Green intellectual capital significantly enhances green innovation and sustainable performance across environmental, social, and economic dimensions among Vietnamese SMEs. Innovation Output	positive	Green innovation and sustainable performance	Reading fidelity high Study strength medium	n=431 0.24
AI applications positively affect the total factor productivity of Chinese service firms, with especially pronounced gains among modern service firms, SMEs, and state-owned enterprises. Firm Productivity	positive	Total factor productivity	Reading fidelity high Study strength medium	n=29 0.24
AI applications enhance enterprise productivity primarily by upgrading human capital through increasing the proportion of highly educated employees while displacing low-skilled workers. Firm Productivity	mixed	Enterprise productivity and workforce skill composition	Reading fidelity high Study strength medium	n=3646 0.24
A one-standard-deviation increase in AI enhances the productivity of Specialized, Refined, Unique, and Innovative SMEs by 6.82%, mediated by labor-structure optimization, innovation stimulation, and management-efficiency improvement. Firm Productivity	positive	SME productivity	Reading fidelity high Study strength medium	n=29 6.82% per standard deviation increase 0.24