Big Data and machine learning are reshaping tourism—improving forecasting, pricing and customer targeting—but firms don’t automatically reap gains: performance depends less on scale or years of tech use than on strategic alignment, data governance and technical readiness.

Main Finding

This 2026 survey (Theodorakopoulos et al., Multidiscip. Sci. J.) integrates peer-reviewed evidence and a structured cross‑regional survey to show that Big Data and Machine Learning (ML) are widely applied across tourism (demand forecasting, dynamic pricing, recommender systems, sentiment analysis) and are perceived to deliver positive economic effects (competitiveness, efficiency, new business formation). However, gains are uneven: technological adoption alone (size or years of use) does not guarantee improved performance — strategic alignment, data governance, skills, and interoperable infrastructure are critical mediators. The paper identifies key technical, organizational, and regulatory bottlenecks and proposes a research/practice agenda emphasizing privacy-preserving learning, interpretability, real-time models, interoperable IoT/streaming architectures, and shared evaluation benchmarks.

Key Points

• Core applications reviewed:
  • Demand forecasting and predictive analytics (regression, tree ensembles, RNN/LSTM, deep learning).
  • Dynamic pricing and revenue management (real-time price optimization).
  • Recommender systems (collaborative, content-based, hybrid, context-aware).
  • Sentiment and opinion mining (SVM/Naive Bayes → CNN/RNN/transformers like BERT).
• Data sources and enabling technologies:
  • Sources: social media and reviews, booking/transactional records, IoT and GPS sensors, mobile apps.
  • Infrastructure/tech stack: Hadoop, Spark (batch & real-time), cloud platforms, real-time streaming, IoT devices, visualization tools.
• Reported economic impacts:
  • Enhanced competitiveness and destination appeal through personalization and targeted marketing.
  • Operational efficiency: staffing, energy and inventory optimization, chatbots, resource allocation.
  • Revenue effects: improved pricing, higher occupancy/revenue per available room, new revenue streams and startups.
  • Labor/market structure: demand for data/AI skills, growth in AI-enabled tourism startups.
• Empirical nuance and heterogeneity:
  • A structured survey finds widespread awareness/adoption but uneven readiness; perceived benefits are common yet not mechanically tied to firm size or tenure with technology.
  • Benefits depend on governance, skills, data integration, model interpretability, and legal/regulatory context.
• Main barriers:
  • Data fragmentation and heterogeneity across sources and jurisdictions.
  • Privacy/security risks and multi‑jurisdictional regulatory complexity.
  • Skill shortages (data scientists, ML engineers) and managerial capacity.
  • Model opacity and lack of domain‑specific interpretable tools for operations.
  • Lack of standardized, open benchmarks and evaluation datasets tailored to tourism.
• Practical recommendations (agenda):
  • Privacy-preserving learning (federated learning, differential privacy) for cross-jurisdictional data use.
  • Interpretable, domain‑specific models for real-time operational decisions.
  • Interoperable architectures bridging IoT, streaming, and transactional systems.
  • Creation and sharing of standardized evaluation datasets/benchmarks for cumulative research.
  • Governance, skills training, and public investment in shared digital infrastructure.

Data & Methods

• Study type: literature survey plus an original structured, cross‑regional survey of tourism organizations; includes case studies (e.g., Barcelona, Amsterdam, Singapore smart-destination deployments).
• Literature mapping: synthesizes peer‑reviewed studies on tasks (forecasting, recommendations, sentiment), data types, model families, evaluation metrics, and reproducibility practices; compares related surveys and gaps.
• Empirical supplement: structured questionnaire administered across regions (details summarized qualitatively in the paper) reporting adoption levels, perceived economic effects, and readiness/capability gaps.
• Analytical approach: qualitative synthesis and taxonomy construction linking data types → tasks → outcomes; tabulated comparisons of technologies and ML techniques; identification of recurring technical/organizational challenges.
• Limitations noted by authors: heterogeneity of primary studies, limited standardized benchmarks and reproducible datasets, and a predominance of descriptive/associational evidence rather than causal impact studies.

Implications for AI Economics

• Productivity and heterogeneous returns:
  • AI/Big Data adoption in tourism can raise productivity and revenues, but returns are heterogeneous and contingent on complementary assets (governance, skills, interoperable data infrastructure). Models of technology adoption should explicitly include these complementarities.
• Measurement and causal inference needs:
  • Current evidence is mostly descriptive and perceived; AI economics requires more causal studies (randomized trials, natural experiments, firm-level panel analysis) to estimate treatment effects of ML systems on productivity, revenues, employment, and welfare.
• Policy and infrastructure:
  • Multi‑jurisdictional privacy rules impede cross-border data integration — policy coordination or privacy-preserving tech (federated learning, DP) are economically important for enabling scalable ML without harming privacy.
  • Public goods investments (shared streaming/IoT platforms, open benchmark datasets) can lower barriers for SMEs and reduce concentration risk from large platform providers.
• Distributional and market-structure concerns:
  • Dynamic pricing and personalized recommendations can increase firm revenue but also raise questions on consumer surplus, fairness, and price discrimination; regulators should monitor welfare and competition effects, especially where platforms exert market power.
• Labor market and skill formation:
  • Demand for data and AI skills in tourism will grow; economic policy should support targeted retraining, education, and incentives to build managerial capacity to capture gains.
• Research agenda for AI economics:
  • Develop standardized outcome metrics for tourism (occupancy-adjusted revenue, destination-level welfare measures).
  • Evaluate privacy‑preserving and interpretable ML methods in field deployments to quantify trade-offs between performance, compliance cost, and adoption.
  • Study spillovers (e.g., how smart-destination infrastructure affects local non-tourism sectors) and startup formation dynamics enabled by Big Data.
  • Incorporate streaming/real-time data externalities and multi-agent interactions (platforms, hotels, transport) into models of market equilibrium.

Concise takeaway: Big Data and ML show clear promise for boosting tourism productivity and economic development, but realizing gains at scale requires complementary investments in governance, interoperable infrastructure, skills, privacy-preserving technologies, and standardized evaluation to move from promising pilots and perceptions to verifiable, equitable economic impact.