Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

MANUSCRIPT REVIEW

Generative AI for Global Market Intelligence, Service Innovation, Customer Experience, and Sustainable Competitiveness in SMEs and Service Firms: A Systematic Literature Review

 

Recommendation: Major revision

 

References “ll.” denote line numbers in the right margin of the reviewed PDF; S1 and S2 are supplementary materials.

 

1. Overall assessment

The manuscript addresses a timely problem and has clear strengths, including a transparent structure, consistent alignment of the results with the five research questions, a detailed extraction matrix, and an honest discussion of the asymmetries in the evidence. However, the work requires substantial revisions.

2. Substantive comments

2.1. Section 2.8 describes a four-domain appraisal (methodological rigor, internal coherence, theoretical clarity, review relevance; high/moderate/limited) and states: “the study-level appraisal was recorded in the extraction matrix (Supplementary Table S1) and summarized in the Results section” (ll. 413–414). However, S1 does not contain any column with these ratings-the only comparable field is “Relevance to SLR,” which concerns relevance to the review, not methodological quality. Table 2 (l. 674) is an aggregate profile of the corpus, not an assessment of individual studies. Nevertheless, S2 marks PRISMA items 11 and 18 as “Complies.” Please add ratings for the four domains for each of the 27 studies and indicate how they affected the weighting of evidence (ll. 405–412), separating methodological quality from relevance.

2.2. Section 2.2 requires a substantive GenAI component (“generative AI, generative artificial intelligence, ChatGPT, large language models (LLMs), AI-generated content, generative chatbots, or closely related GenAI architectures,” ll. 164–166) and an SME/service-firm context; mixed populations are permitted only when SME-specific evidence can be isolated (ll. 169–170). The authors’ own coding in S1 identifies cases that require justification: Necula and Rieder [25]-“Type of generative AI: Variational autoencoder plus adaptive AI”; Gunduz et al. [40]-“Transformer embeddings”; Zhao et al. [4]-unit of analysis: 337 cities (l. 634), GenAI: “Multimodal LLM-assisted sentiment scoring”; Metwally et al. [17]-“Firm size: Mixed; not restricted to SMEs.” Since 27 full texts were excluded because “GenAI was not a central phenomenon,” and 30 because of the absence of an SME/service context (ll. 559–561), the authors should define the conceptual boundary of GenAI (LLM-based systems or generative models more broadly) and, for the studies listed, justify their inclusion or exclude them and recalculate the synthesis.

2.3. The third search block (ll. 227–235) contains only market- and innovation-related terms; it does not include trust, transparency, privacy, bias, governance, responsible AI, AI literacy, or readiness. Therefore, the conclusion that the evidence for RQ5 is “incipient and fragmented” (Table 3) may partly reflect search sensitivity rather than the state of the literature. The search should be expanded, or this limitation should be explicitly acknowledged in Section 4.5. The 2024–2026 window omits 2023-the first full year after the release of ChatGPT (November 2022)-with only a general justification (ll. 179–181), and 2026 is covered only through August 15 (l. 207). The manuscript does not mention citation searching (snowballing) or sources other than Scopus and WoS; this should be identified as a limitation regarding completeness.

2.4. “Subject-area filters were applied, where appropriate” (l. 250) and “Relevant Web of Science categories were applied” (ll. 265–267) do not allow the search to be reproduced; please list the categories applied for both databases. Figure 1 includes a stage titled “Records identified after subject area/category and document type filters (n = 860),” although the sum of records from both databases before this stage is identical (536+324); this requires explanation.

2.5. The sentence “In the working record provided, the 116 full-text exclusions were provisionally distributed…” (ll. 558–559) suggests a provisional classification. S2 considers PRISMA item 16b to be met, although only aggregate numbers are provided. Please provide a list of the 116 excluded publications with the individual reason for each and remove the word “provisionally.” Given seven authors and two reviewers (ll. 290–292), the manuscript does not report a measure of inter-reviewer agreement; this should be added if it was calculated.

2.6. The labels in Table 3 (“Moderately strong,” “Moderate,” “Incipient and fragmented”) lack decision rules analogous to those in Table 1; Section 2.12 provides only general principles. The assessment of RQ1 as “moderately strong” (ll. 977–983) is based mainly on computational studies, which the authors themselves note are “more consistent in demonstrating that GenAI can perform certain analytical tasks than in establishing generalizable organizational relationships” (ll. 668–670). Please formulate criteria or soften the labels.

2.7. The conclusion “GenAI is emerging as an organizational capability” (l. 1351) is too categorical and conceptually ambiguous-access to technology is not an organizational capability, as the authors themselves argue in Section 4.2 (ll. 1084–1090). Similarly, “sustainable competitiveness” in the title and conclusions is not supported by longitudinal evidence, which the authors themselves rate as “Very limited” (Table 2) and “scarce” (ll. 924–926). The conclusions should distinguish task-level benefits from enduring firm-level effects. Minor point: the result “satisfaction from 3.2 to 4.6” (l. 741) requires the scale to be specified.

2.8. The “Country/region” column in S1 conflates data location, firm location, and author affiliations (e.g., “Turkey; global Kickstarter data”; “Global social-media data; Australia/Saudi Arabia affiliations”; “United States retail data; Romania-based study”). The description in Section 3.2 (ll. 577–585) lists China, Japan, Vietnam, the Philippines, India, the United Kingdom, Ireland, Germany, the United States, Canada, and Ecuador, while omitting, among others, Australia and Hong Kong, which are already described as “particularly visible” in Section 3.6 (ll. 866–867). Please define the country coding rule and harmonize the description-this is important given the word “Global” in the title.

3. Formal and linguistic comments

References: item 20-year 2024 with a DOI containing “025” (10.1007/s10843-025-00385-w; ll. 1437–1439), to be verified; item 14-“In Proceedings of the Proceedings of…” (ll. 1422–1423); items 26 and 39-series names (Lecture Notes in Mechanical Engineering; Communications in Computer and Information Science) recorded as proceedings titles; item 30-distorted publisher data (“John Wiley & Sons, Ed.; Blackwell, 2008;”); item 34-missing publisher. In Section 2.2, the author–year citation “Petticrew and Roberts (2008)” (l. 158) appears alongside the numeric style used throughout the work.

The language is understandable but occasionally unnatural: “Latin America was represented much more limitedly” (l. 584; better: was much less represented); inconsistency between “an RAG chatbot” (ll. 640, 741–742) and “a RAG architecture” (l. 1023)-when the abbreviation is pronounced as a word, “a RAG” is correct; the term “phygital” (l. 740) appears in the text without definition.

However, before acceptance, the following are necessary: documentation of the quality assessment for each study (2.1), clarification of the GenAI definition and verification of the included studies (2.2), completion of the search strategy and filters (2.3–2.4), an auditable exclusion list (2.5), operationalization of evidence strength (2.6), softening of the conclusions (2.7), harmonization of the geographic profile (2.8), and bibliographic and linguistic correction (3).

 

 

Author Response

See attached Word file.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Please see the attached review for detailed comments and suggestions.

Comments for author File: Comments.pdf

Author Response

See attached Word file.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript possesses clear potential, featuring several notable strengths: the timeliness and high relevance of the topic, a clear overall structure, the application of the PRISMA guidelines, and a generally prudent approach to the conclusions.

However, several methodological, analytical, and reporting issues require significant attention. These concerns prevent a confident assessment of the conclusions in their current form and necessitate a major revision before the manuscript can be reconsidered for publication.

Major Revisions Required

1. Search Strategy and Research Questions Alignment The search strategy must be explicitly aligned with the five research questions. Currently, the query lacks sufficient terms covering key dimensions such as organizational capabilities, AI readiness, AI literacy, governance, data privacy, algorithmic bias, transparency, and human oversight. Please expand the search strings and explain how the updated terms capture these crucial domains.

2. Quality Assessment Matrix The manuscript announces four quality assessments conducted for each of the 27 included studies, yet these detailed valuations are missing from the supplementary matrix. Please integrate these evaluations comprehensively into the supplementary materials or the main text.

3. Exclusion Transparency Please provide an individual listing of all studies excluded during the full-text screening stage, along with their specific corresponding reasons for exclusion. Furthermore, the distribution and reporting of the 116 exclusions should be treated as definitive rather than described as "provisional."

4. Categorization and Grading Criteria Please establish and present explicit criteria used to classify study results (e.g., positive, conditional, exploratory, or mixed) and to assign qualitative levels or maturity tags such as "moderate" or "fragmented."

5. Conceptual Mapping Table We strongly recommend including a synthesis table that maps each component of the conceptual model directly to the supporting studies, their respective methodological designs, and their quality appraisal scores.

6. Geographical and Sectoral Generalization The manuscript claims a "global" scope and generalizes its findings to general service enterprises. However, the underlying evidence appears primarily concentrated in tourism, hospitality, and retail. Please provide a robust justification for this broader generalization or narrow the study's scope accordingly.

7. Temporal Scope and 2026 Data Please justify the selection of the 2024–2026 timeframe. In particular, note that 2026 is an ongoing and incomplete year; direct numerical comparisons between 2026 and full preceding years are methodologically misleading and should be adjusted or contextualized.

Minor Revisions and Technical Corrections

• Figure 2: Please improve the legibility, resolution, and labeling of Figure 2.

• Author Contributions: The author contributions section still contains placeholder template markers (such as "X.X." and "Y.Y.") and requires careful proofreading to correct minor spacing and punctuation errors.

Conclusion

The manuscript addresses an exciting and timely topic with a sound foundational structure. Nevertheless, addressing the methodological gaps concerning search reproducibility, missing quality appraisal data, and clearer categorization criteria is essential. Consequently, a major revision is required before we can recommend the manuscript further.

Author Response

See attached Word file.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I would like to thank the Authors very much for their response and the revisions made to the manuscript.

Author Response

Thank you for your valuable comments.

Reviewer 2 Report

Comments and Suggestions for Authors

Please see the attached review for detailed comments and suggestions.

Comments for author File: Comments.pdf

Author Response

see report

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for the substantial revision. The study-level quality assessments, full-text exclusion list, evidence-coding definitions, conceptual mapping table, and limits on geographic and temporal interpretation address most of my previous concerns. I have three remaining requests:

1. Please enlarge or simplify the labels in Figure 2 so they remain legible at normal page size.
2. Please correct the apparent typo in the last row of Table 4 (p. 29): the score for study [17] appears as “.00”, whereas it appears as “1.00” elsewhere in the table.
3. In Supplementary Table S1, the Scopus/WoS provenance field is recorded as “NR” for all 27 included studies. Please identify the source database for each study where this can be reconstructed, or explain why that information is unavailable.

The original search was not expanded to include the requested RQ4/RQ5 terms. I consider the authors’ explicit acknowledgment of this limitation and their narrower interpretation of those findings an acceptable response, provided that qualification is retained throughout the final manuscript.

Author Response

see RESPONSE (WORD)

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for the careful revision. Both of my previous concerns have been addressed. One minor documentation issue remains in the supplementary material. Supplementary Table S1 still contains several working annotations such as pending verification/recheck notes, including “dual-reviewer confirmation pending” in the Methodological Quality worksheet. These should be resolved or removed before publication. In addition, Supplementary Table S2 should be updated so that the PRISMA checklist points to the current Methodological Quality worksheet and uses the revised section titles. These are documentation issues only and require no further analysis.

Author Response

Reviewer 2 – Comment

Thank you for the careful revision. Both of my previous concerns have been addressed. One minor documentation issue remains in the supplementary material. Supplementary Table S1 still contains several working annotations such as pending verification/recheck notes, including “dual-reviewer confirmation pending” in the Methodological Quality worksheet. These should be resolved or removed before publication. In addition, Supplementary Table S2 should be updated so that the PRISMA checklist points to the current Methodological Quality worksheet and uses the revised section titles. These are documentation issues only and require no further analysis.

Authors’ Response:

We sincerely thank the reviewer for acknowledging that the previous concerns have been satisfactorily addressed and for identifying these remaining documentation issues.

We have carefully reviewed and cleaned Supplementary Table S1, removing internal working annotations, pending verification and recheck notes, and provisional statements, including “dual-reviewer confirmation pending” in the Methodological Quality worksheet. The supplementary material now reflects the finalized documentation of the 27 included studies, while preserving the original methodological assessments, quantitative results, and supporting evidence.

We have also updated Supplementary Table S2 to ensure consistency with the revised manuscript and the final version of Supplementary Table S1. Specifically, PRISMA checklist items 11 and 18 now explicitly reference the Methodological Quality worksheet, where the study-level assessments are documented. Additionally, the checklist references and section titles have been aligned with the current manuscript, including Section 2.12 (Assessment of Evidence Consistency and Applicability) and Section 3.8 (Consistency and Scope of the Evidence).

These revisions are exclusively documentary and do not involve additional analyses or modifications to the study findings. Both supplementary tables have been updated accordingly for the final submission.