Exploring the Role of AI and Software Solutions in Shaping Tourism Outcomes: A Factor, Neural Network, and Cluster Analysis Across Europe

Abstract

Tourism and digitalization have become increasingly interconnected, yet the complex, nonlinear relationships between technological adoption and tourism performance remain underexplored. This study aims to examine how enterprise software solutions influence tourism indicators across European countries. Using a triangulated methodological approach, we employed factor analysis to identify underlying dimensions, neural network modeling to detect nonlinear relationships, and hierarchical clustering to group countries based on digital and tourism profiles. The results consistently highlight CRM (Customer Relationship Management) as the most influential technological factor linked to both the net occupancy rate of beds and the number of nights spent at tourist accommodations. While AI (artificial intelligence) technologies currently have less impact, their importance is growing, as seen in emerging patterns. Cluster analysis further confirms that countries with higher CRM adoption tend to cluster together and show better tourism performance, indicating a clear connection between digital maturity and sector competitiveness. These findings emphasize the strategic importance of CRM as a transformative tool in hospitality and tourism management, while also recognizing the potential of AI to shape future trends. The study offers empirical support for tailored digital policies across European regions to promote inclusive and sustainable tourism growth.

1. Introduction

In recent years, the digital transformation of the tourism industry has become a crucial factor for competitiveness, innovation, and organizational resilience. As new technologies more and more influence the operational and strategic aspects of hospitality businesses, researchers and policymakers have shifted their attention to the effects of this change on economic performance, sustainability, and consumer engagement [1,2,3]. Among these new technologies, artificial intelligence (AI) stands out because of its significant disruptive potential, thanks to its ability to surpass the limitations of traditional information systems. From automated service customization and dynamic pricing to predictive analytics and real-time operational improvements, AI-driven tools are transforming how tourism services are created, provided, and experienced [4,5,6].

The academic literature increasingly reflects this trend. Landmark studies, such as those conducted by Tussyadiah [7] and Ivanov et al. [8], emphasize the transformative role of AI and robotics, especially in improving customer interactions and operational efficiency. Similarly, Buhalis and Amaranggana [9], along with Grundner and Neuhofer [10], highlight the ongoing shift toward smart tourism ecosystems, where data-driven decisions and automated, learning-based systems are becoming standard. Recent contributions, including that of Agiropoulos et al. [11], further expand this view by analyzing the digitalization gap across European countries and its impact on tourism competitiveness. Despite these insights, a significant gap remains in the empirical literature: few studies have evaluated the combined effects of multiple digital solutions, particularly CRM (Customer Relationship Management), ERP (Enterprise Resource Planning), BI (Business Intelligence), and AI, on tourism performance at the national level.

This research aims to address that gap by examining how adopting advanced digital solutions in hospitality businesses relates to improved tourism performance. The study focuses on key software tools such as Customer Relationship Management (CRM), Enterprise Resource Planning (ERP), Business Intelligence (BI), and AI-based technologies. It evaluates their impact on two leading indicators: the net occupancy rate of bed-places (NORB) and the number of nights spent at tourist accommodations per thousand inhabitants (NSTA). To analyze these relationships, the research employs a solid methodological framework that combines factor analysis, Multilayer Perceptron (MLP) neural networks, and hierarchical cluster analysis.

The study’s originality lies in its integrative and comparative approach, based on harmonized EU-level data provided by Eurostat. Unlike previous studies, which often focus on case-based analyses or evaluate only a single technology type, this research offers a comprehensive view of how digitalization influences tourism performance across Europe. The novelty of this contribution is twofold: it statistically assesses the individual impact of each technology and identifies national typologies through segmentation methods, thereby revealing structural differences between countries. Additionally, the use of neural networks introduces a sophisticated modeling approach capable of capturing nonlinear dynamics that traditional statistical methods might miss.

The structure of this paper is as follows: the introductory section is followed by a literature review on digitalization in tourism and its impact on performance. Next, the data, variables, and methods are presented, followed by the section on empirical results. The final two sections provide interpretative discussions and general conclusions, including implications for public policy and future research directions.

2. Literature Review and Hypothesis Development

2.1. Artificial Intelligence and Systemic Transformations in European Tourism

Within the European context, where tourism is experiencing a major digital transformation, artificial intelligence (AI) no longer acts just as a technological addition to traditional methods; it has become a key driver in redefining the tourism value chain [12,13,14]. By integrating artificial neural networks (ANNs) and other advanced digital tools, many authors see a tourism industry where personalization, efficiency, and predictive analytics set new standards [7,15].

This transformation is supported by studies like Siddik et al. [16], which emphasize AI’s role in tourism growth across ten of the most visited countries from 2010 to 2022. Using a combination of ANN models and traditional econometric analysis, these researchers show that AI, along with GDP, is a key factor driving tourism demand. Other structural variables, such as GDP, FDI, and urbanization, also impact tourism. While GDP directly increases demand, inflation has the opposite effect [17,18,19,20,21]. The impact of FDI remains mixed, and the effects of urbanization largely depend on the specific context [22,23].

Within this theoretical framework, AI emerges not just as an automation tool but as a transformative agent that enables tourism to continuously adapt to external pressures, from macroeconomic fluctuations to increasingly sophisticated traveler expectations [6].

2.2. AI, CRM, and Performance in the Accommodation Sector

Digital transformation in tourism is most evident in how accommodation businesses adopt software solutions, especially those involving AI and CRM. The use of these technologies goes beyond improving internal efficiency, aiming to enhance customer experience and provide personalized service. In fact, CRM systems, when combined with ERP and BI platforms, allow detailed analysis of tourist preferences and more effective management of interactions throughout the decision-making process.

The range of AI technologies, from chatbots and service robots to virtual and augmented reality, greatly improves service quality while also raising the expectations of modern travelers [24,25,26,27,28,29,30,31,32,33,34,35]. These tools enable continuous, smooth, and context-aware interactions, strengthening customer loyalty and increasing destination appeal [10,36]. For tourism professionals, this change means fewer repetitive tasks and more focus on improving the experiential aspects of services [37,38].

Still, the literature does not avoid acknowledging the challenges. Researchers often raise concerns about privacy, transparency issues, and social alienation risks [39,40]. Emotional reactions to AI, from excitement to discomfort, underline the complex social aspects of AI acceptance, heavily influenced by demographic factors like age, education, and socioeconomic status [41].

Beyond its operational benefits, AI is becoming increasingly vital in tourism forecasting, where machine learning and deep learning algorithms now outperform traditional statistical models [42,43,44]. These methods have proven successful not only in predicting demand but also in analyzing customer emotions and behavior [45,46,47].

Among all technological applications, CRM is the most directly linked to performance in the accommodation industry. Its ability to gather valuable data on tourist preferences, behavioral history, and previous interactions enables businesses to deliver high-value services. This, in turn, positively influences both bed occupancy rates and the number of nights guests stay at lodging establishments [1,48].

Building on these empirical findings and the underlying theoretical framework, we establish the study’s first hypothesis:

Hypothesis H1.

Among the software solutions used by enterprises in the accommodation sector (AI technologies, ERP, CRM, BI), CRM software shows the strongest link to the net occupancy rate of bed-places and the number of nights spent at tourist accommodation establishments, while AI technologies are becoming increasingly relevant.

2.3. Digitalization, Performance, and Tourism Clustering in the EU

Recent scholarship in tourism studies increasingly emphasizes regional disparities in digital technology adoption and tourism performance across Europe. Agiropoulos et al. [11] introduce a rigorous methodology based on unsupervised machine learning, which allows for the identification of meaningful regional typologies. By combining PCA, K-means clustering, and exploratory data analysis (EDA), they show that the success of regional tourism cannot be fully understood without an integrated view of economic, cultural, and infrastructural variables [49,50].

Research shows that EU member states can be grouped based on a consistent set of indicators, such as occupancy rates, the number of overnight stays, and the level of digitalization among tourism enterprises [51,52]. These groups not only mirror regional competitiveness but also offer a practical foundation for tailored public policy approaches that address local needs [53,54].

Within these typologies, the digital dimension plays a decisive role. Regions with high levels of CRM and AI adoption tend to outperform others in terms of infrastructure, innovation capacity, and tourism performance [2,55]. AI does more than optimize individual service delivery; it also contributes to shaping distinct regional models for tourism development [11,56].

Building on these contributions, the second hypothesis of this study states:

Hypothesis H2.

European Union countries can be grouped based on the software solutions used by enterprises in the accommodation sector and on key accommodation performance indicators, such as the net occupancy rate of bed-places and nights spent at tourist accommodation establishments.

3. Materials and Methods

3.1. Research Design

This study uses a quantitative approach with both exploratory and confirmatory goals to better understand how adopting digital technologies, particularly those based on artificial intelligence, affects the performance of the tourism sector across Europe. The research is centered around two testable hypotheses, based on the previously outlined theoretical framework, which highlights the increasing influence of emerging technologies in tourism [7,10,11]. The analysis focuses on two main aspects: first, examining the relationship between the use of specific software tools and tourism performance; second, exploring national profiles shaped by these variables. Choosing a quantitative strategy is driven by the need to identify significant correlations and create reliable models with both academic and practical importance.

The analysis is based on Eurostat data for the year 2023, and the study follows a cross-sectional design, providing a comparative overview of digital adoption and tourism performance across European Union member states during this period.

3.2. Selected Variables

To analyze the relationship between digitalization and tourism performance in Europe, the study uses a specific set of variables organized into two main categories: the level of digital transformation in the hospitality industry and the efficiency of tourism at the national level. All data are sourced from official Eurostat sources, ensuring validity and comparability across countries.

The digitalization dimension is evaluated based on how much advanced technological solutions have been integrated into the daily operations of hospitality businesses. A key measure here is the use of AI technologies, indicated by the percentage of companies that have adopted at least one AI solution into their operations. These solutions include various components such as machine learning, image recognition, natural language processing, voice recognition, automated planning, reasoning systems, and automated decision-making support. This variable provides a broad overview of how the hospitality industry is embracing emerging technologies that can dramatically transform operational procedures and customer engagement.

In addition to AI, the study covers other digitalization forms with significant operational effects. The use of ERP systems shows an organization’s ability to integrate and manage internal information flows, from resource management to accounting and logistics. The implementation of CRM solutions reflects a strategic focus on automating and personalizing customer interactions, which directly improves the tourist experience and builds customer loyalty. Similarly, the adoption of BI tools highlights an organizational emphasis on advanced data analysis and strategic decision-making based on reliable, consolidated insights.

Regarding tourism performance, the study concentrates on two indicators that provide complementary insights. The Net Occupancy Rate of Bed Places measures how effectively available infrastructure is used, serving as a traditional indicator of a destination’s attractiveness and seasonal demand patterns. Meanwhile, the Number of Nights Spent in Tourist Accommodations per 1000 Inhabitants offers a contextualized perspective of tourism intensity relative to the population size, reflecting a destination’s capacity to attract and hold visitors over time.

Table 1. Research variables.

Variable	Dataset	Measures	References
AITH	Enterprises use at least one of the AI technologies	Percentage of enterprises	[57]
ERP	Enterprises that have an ERP software package to share information between different functional areas	Percentage of enterprises	[58]
CRM	Enterprises using Customer Relationship Management (CRM) software	Percentage of enterprises	[58]
BI	Enterprises using Business Intelligence (BI) software	Percentage of enterprises	[58]
NORB	Net occupancy rate of bed-places in hotels and similar accommodation	Percentage	[59]
NSTA	Nights spent at tourist accommodation establishments	Nights spent per thousand inhabitants	[60]

Source: developed by the authors based on Eurostat [57,58,59,60].

presents the variables chosen for analysis in this study.

Through selecting these variables, the study aims to build an integrated analytical framework that links emerging digital technologies with the dynamics and actual performance of the European tourism sector. Such an approach enables not only a thorough evaluation of the direct impacts of digitalization but also a systematic identification of structural differences across countries in terms of technological maturity and tourism competitiveness.

Figure 1 illustrates the overall research process and analytical workflow applied in this study.

3.3. Methods

To gain a deeper understanding of the relationship between the adoption of digital technologies and the performance of the European tourism sector, the study combined advanced quantitative methods chosen both for their exploratory value and for their ability to provide robust empirical validation of the hypotheses. The methodological design unfolded along three complementary analytical directions: factor analysis, artificial neural networks analysis, and hierarchical cluster analysis. Each method served a distinct purpose, yet they converged toward a more comprehensive understanding of the phenomenon under investigation.

Figure 2 presents the employed methods to test the hypotheses and obtain the empirical analysis results, outlining the sequential steps from data collection to statistical validation.

The first stage involved exploratory factor analysis, which simplified the variables and identified underlying dimensions that might indicate a common structure between digitalization levels and tourism performance. This method used the principal component approach, retaining only one factor, meaning that only components with eigenvalues greater than one were extracted [61,62].

The formula for factor analysis is (1):

$$X=LF + \in$$

(1)

$X$—observed variables.

$L$—matrix of factor loadings.

$F$—latent factors.

$\in$—errors.

Before conducting factor analysis, the adequacy was evaluated. The Kaiser-Meyer-Olkin (KMO) measure indicated a good fit for dimensional reduction, while Bartlett’s test of sphericity confirmed significant correlations among variables (p < 0.001).

To capture potentially nonlinear relationships between explanatory and outcome variables, the research used artificial neural networks with a Multilayer Perceptron (MLP) architecture using SPSS Statistics for Windows, Version 27.0 (IBM Corporation, Armonk, NY, USA). This approach proved highly effective for analyzing complex interactions [63,64] that linear regression methods often miss, especially in a multi-dimensional context like digitalization and tourism. The MLP model was designed with an input layer that included four digital variables, AI, ERP, CRM, and BI, followed by a hidden layer with a sigmoid activation function to capture nonlinearities, and an output layer estimating two tourism performance indicators.

The network architecture typically comprises an input layer, one or more hidden layers, and an output layer (2):

$$y=({\sum }_{i=1}^n{w}_i{x}_i+b)=\phi (W^{T}X+b)$$

(2)

y—vectors of outputs.

w, x—vectors of weights and inputs.

b—bias.

i—cases.

φ—activation functions.

The activation function used in the hidden layers was the sigmoid function, defined as (3):

$$f\left(n\right)={\displaystyle \frac{1}{1+e^{-n}}}$$

(3)

n—input variables.

f(n)—output variables.

The network parameters were automatically calibrated during training, while the model was validated by dividing the sample into training and testing sets.

In the final stage of analysis, hierarchical cluster analysis was used to identify distinct national typologies based on digitalization levels and tourism performance, employing Ward’s method [65,66].

The function employed in Ward’s algorithm is expressed as [66] (4):

$$\Delta \left(\mathrm{A},\mathrm{B}\right)=\sum _{i\in A\cup B}{\left|\left|{\overrightarrow{x}}_i-{\overrightarrow{m}}_{A\cup B}\right|\right|}^2-\sum _{i\in A}{\left|\left|{\overrightarrow{x}}_i-{\overrightarrow{m}}_A\right|\right|}^2-\sum _{i\in B}{\left|\left|{\overrightarrow{x}}_i-{\overrightarrow{m}}_B\right|\right|}^2={\displaystyle \frac{n_A{n}_B}{n_A{+n}_B}}{\left|\right|{\overrightarrow{m}}_A-{\overrightarrow{m}}_B\left|\right|}^2$$

(4)

${\overrightarrow{m}}_{\mathrm{j}}$—the center of cluster j.

${\mathrm{n}}_{\mathrm{j}}$—number of points in cluster j.

Δ—merging cost of combining the clusters A and B.

i—cases.

This technique grouped European countries based on the similarity of their digital and tourism profiles by minimizing variance within each group. Distances between observations were calculated using the Euclidean metric, and group formation aimed to reduce total intra-cluster variance. The results were visually represented in a dendrogram that showed meaningful clustering patterns and enabled interpretation of differences among member states regarding both the adoption of digital technologies and tourism performance.

By combining three methods, dimensionality reduction through factor analysis, relationship modeling with neural networks, and European space segmentation via cluster analysis, the study presents a clear, detailed, and statistically supported picture of how digital technologies influence tourism performance. This complex methodological approach not only tests the initial hypotheses but also provides insights valuable for public policy creation and strategy planning in a digitally driven European tourism sector.

4. Results

The results of the factor analysis provide valuable insights into the relationships between digital technologies and tourism performance indicators in the accommodation sector. The correlation matrix acts as a basic foundation, showing how the six variables studied are interconnected (

Table 2. Correlation Matrix, KMO and Bartlett’s Test.

		AITH	ERP	CRM	BI	NORB	NSTA
Correlation	AITH	1.000	0.167	0.296	0.377	0.430	0.357
	ERP	0.167	1.000	0.700	0.458	0.620	0.358
	CRM	0.296	0.700	1.000	0.788	0.673	0.307
	BI	0.377	0.458	0.788	1.000	0.546	0.397
	NORB	0.430	0.620	0.673	0.546	1.000	0.510
	NSTA	0.357	0.358	0.307	0.397	0.510	1.000
Sig. (1-tailed)	AITH		0.203	0.067	0.026	0.013	0.034
	ERP	0.203		0.000	0.008	0.000	0.033
	CRM	0.067	0.000		0.000	0.000	0.060
	BI	0.026	0.008	0.000		0.002	0.020
	NORB	0.013	0.000	0.000	0.002		0.003
	NSTA	0.034	0.033	0.060	0.020	0.003
Kaiser-Meyer-Olkin Measure of Sampling Adequacy.							0.719
Bartlett’s Test of Sphericity		Approx. Chi-Square					71.873
		df					15
		Sig.					0.000

Source: author’s design using SPSS v.27.0 (IBM Corporation, Armonk, NY, USA).

).

Customer Relationship Management (CRM) software is strongly linked to both the net occupancy rate of bed-places (r = 0.673) and the use of Enterprise Resource Planning (ERP) systems (r = 0.700), as well as Business Intelligence (BI) tools (r = 0.788). This finding indicates that CRM not only closely aligns with other technological solutions but also correlates with key performance indicators in the tourism sector. Moreover, AI technologies, while showing more modest correlations across the matrix, still maintain meaningful associations, particularly with the net occupancy rate (r = 0.430) and nights spent at accommodation establishments (r = 0.357). Although these correlations are lower than those of CRM or ERP, they suggest an emerging, though not yet dominant, role for AI in shaping enterprise outcomes in this area.

The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy, with a value of 0.719, indicates that the dataset is suitable for factor analysis, exceeding the commonly accepted threshold of 0.6. This result suggests that the correlation patterns are sufficiently compact and allow for reliable factor extraction. Bartlett’s Test of Sphericity further confirms the appropriateness of proceeding with factor analysis, with a highly significant result (Chi-square = 71.873, p < 0.001), implying that the variables share enough common variance to justify dimensional reduction.

The communalities table shows how much variance in each variable is explained by the extracted factor (

Table 3. Communalities and factor matrix.

	Initial	Extraction	Factor1
AITH	0.260	0.186	0.432
ERP	0.579	0.486	0.697
CRM	0.798	0.769	0.877
BI	0.685	0.587	0.766
NORB	0.608	0.677	0.823
NSTA	0.354	0.267	0.516

Source: author’s design using SPSS v.27.0 (IBM Corporation, Armonk, NY, USA).

).

CRM, with a high communality of 0.769, appears as the most strongly explained variable by the factor solution. This result indicates that nearly 77% of the variance in CRM use among companies is shared with the other variables in the model. Similarly, the communalities for ERP (0.486), BI (0.587), and NORB (0.677) are moderate to high, while AITH and NSTA have lower communalities at 0.186 and 0.267, respectively. This result again suggests that AI technologies, although present in the structural model, contribute less shared variance and are not yet fully integrated into the dominant pattern of digital engagement within the accommodation sector.

The factor matrix shows a single extracted factor that accounts for the shared variance among the variables. The highest loadings are seen for CRM (0.877), NORB (0.823), and BI (0.766), indicating that these variables are most central to the latent dimension identified by the analysis. ERP also loads strongly on this factor (0.697), confirming its structural importance. AI technologies, however, have a comparatively weak loading of 0.432, while NSTA, although conceptually relevant, loads only moderately (0.516). This outcome supports the idea that CRM, BI, and ERP are the leading digital enablers in this ecosystem, while AI remains peripheral, possibly due to lower adoption rates or a delay in observable impact.

Taken together, these results provide empirical support for evaluating the proposed hypothesis H1. The hypothesis suggests that among software solutions used by enterprises in the accommodation sector, CRM software shows the strongest link to performance indicators such as bed occupancy rates and nights spent, while AI technologies are increasingly relevant. The empirical evidence from the factor analysis clearly supports this idea. CRM not only exhibits the strongest correlation with NORB but also has the highest communality and factor loading, making it the most influential variable in the shared technological-performance space. The strong correlations and factor loadings of BI and ERP further place CRM within a technologically consistent cluster of high-impact tools.

Regarding AI technologies, their relatively lower factor loadings and communalities indicate that they are still in the early stages of integration and influence. However, their statistically significant correlations with performance indicators like NORB and NSTA suggest a path toward increasing relevance. These patterns imply a future where AI could become more deeply woven into the operational and strategic framework of tourism businesses, especially as data-driven personalization, automation, and forecasting grow more central to customer experience and operational efficiency.

In conclusion, the data support hypothesis H1. CRM software is clearly the most strongly linked to key performance indicators in the accommodation industry, while AI technologies, although not yet as influential, show promising signs of growing importance. This finding indicates a dual path: strengthening of current software solutions and gradual growth of AI’s role in shaping tourism business results.

The analysis conducted using a Multilayer Perceptron (MLP) artificial neural network reveals complex, nonlinear relationships among the technological variables studied (AITH, CRM, BI, and ERP) and key tourism performance indicators (NORB and NSTA). This modeling approach, more adaptable than linear methods, helps uncover hidden dynamics that traditional regression techniques might miss.

The neural network employed in the study was constructed with four input variables: AITH, ERP, CRM, and BI, all rescaled using the adjusted normalization method to ensure comparability across variables with different scales. The model contained one hidden layer with a single processing unit, employing a sigmoid activation function to capture nonlinear interactions between inputs and outputs. The output layer comprised two dependent variables: NORB and NSTA (Figure 3).

Training was performed using a standard backpropagation algorithm, with the stopping criterion defined as one consecutive step without a decrease in the sum of squared errors, preventing overfitting and optimizing convergence. The training phase achieved a Sum of Squares Error (SSE) of 0.632 and an average overall relative error of 0.678, with variable-specific relative errors of 0.505 for NORB and 0.886 for NSTA. The testing phase confirmed the model’s generalizability, yielding an SSE of 0.344 and an average overall relative error of 0.825, with respective relative errors of 0.315 for NORB and 1.045 for NSTA. Training time was negligible, reflecting the model’s computational efficiency given the moderate data size and network complexity.

Table 4. Parameter estimates.

Parameter Estimates
		Hidden Layer 1	Output Layer		Importance	Normalized Importance
		H (1:1)	NORB	NSTA
Input Layer	(Bias)	0.544
	AITH	0.970			0.220	51.9%
	ERP	1.094			0.249	58.6%
	CRM	1.857			0.424	100.0%
	BI	0.471			0.107	25.1%
Hidden Layer 1	(Bias)		−0.908	−1.631
	H (1:1)		2.275	0.994

Source: authors’ design using SPSS v.27 (IBM Corporation, Armonk, NY, USA).

presents the estimated parameters of the model.

The internal structure of the network reveals these dynamics. In the hidden layer, the CRM input node has the highest weight (1.857), significantly exceeding ERP (1.094), AI technologies (AITH, 0.970), and BI tools (0.471). These weights are then transmitted through the hidden node, which connects to both output variables with different weights: a strong 2.275 to NORB and a moderate 0.994 to NSTA. The bias terms are also notable, especially the negative biases at the output layer, which indicate threshold shifts in activation functions, further influencing how the model interprets input combinations.

The most substantial evidence supporting the hypothesis comes from the normalized importance values. CRM stands out as the most influential predictor by a large margin, with a normalized importance score of 100%. ERP and AI technologies follow at some distance, with scores of 58.6% and 51.9%, respectively, while BI remains behind at 25.1%. This ranking of influence confirms the unique and outsized role that CRM plays in shaping enterprise performance within the accommodation sector, especially regarding NORB. The model clearly demonstrates CRM’s ability to organize customer engagement in ways that directly impact occupancy rates and, to a lesser extent, the length of stays.

However, the findings also support the secondary part of hypothesis H1 regarding the growing significance of AI technologies. Although AI does not have as much influence as CRM or ERP, its relative importance (over 50%) remains considerable, especially since artificial intelligence in tourism is still an emerging field. The significant weight given to AITH in the hidden layer highlights this trend, indicating that AI technologies are starting to become a meaningful part of operational decision-making.

Based on these results, hypothesis H1 is well supported. The neural network model confirms that CRM software shows the strongest link to the primary performance metric, the net occupancy rate of bed-places, while also playing a minor role in the number of nights spent. At the same time, AI technologies, though currently less influential, are becoming more relevant and have the potential to become a key driver soon. What emerges is a layered digital ecosystem where CRM remains the most mature and impactful technology, while AI offers promise for future growth and integration.

The hierarchical clustering analysis using Ward’s method provides a detailed perspective on the variation among European countries regarding digital adoption and tourism performance. By grouping countries based on similarities in their use of AI technologies, enterprise software (ERP, CRM, BI), and key tourism metrics, namely the net occupancy rate of bed-places (NORB) and the number of nights spent at tourist accommodations (NSTA), the analysis uncovers distinct structural patterns that explain how technology and tourism interact across the continent (Figure 4 and

Table A1. Clusters data.

	AITH	ERP	CRM	BI	NORB	NSTA
Czechia	2.63	18.43	17.27	6.29	43.4	5157.45
Germany	5.38	30.49	35.63	8.35	60.17	5190.65
Luxembourg	4.87	21.29	22.77	9.48	42.74	5275.28
Estonia	1.64	32.03	30.58	13.78	48	4667.21
Bulgaria	0.76	19.09	14.11	6.90	55.9	4166.6
Finland	2.46	36.45	61.76	47.86	51.8	4103.56
Belgium	7.98	27.36	26.48	22.53	59.57	3806.3
Lithuania	0.55	31.47	26.57	8.39	57.2	2965.7
Hungary	2.46	38.62	30.01	8.54	49.2	3175.95
Poland	1.65	30.40	32.12	8.33	50.7	2524.86
Slovakia	2.31	16.61	12.19	6.83	42.95	2645.13
Latvia	1.01	41.45	18.91	9.78	39.9	2321.21
Romania	0.06	11.90	8.54	1.35	29.14	1532.73
Cluster A means	2.60	27.35	25.92	12.19	48.51	3656.36
Netherlands	5.91	34.34	46.28	21.28	65.5	7987.77
Portugal	5.26	39.45	34.40	27.02	61.6	8071.68
Italy	3.99	38.30	42.47	30.68	54.5	7579.51
Slovenia	6.53	20.08	13.97	4.41	47.74	7606.53
Ireland	2.68	32.27	41.10	25.65	76	7708.8
Denmark	6.60	31.58	28.79	23.39	60	6564.7
France	2.47	43.94	35.28	10.52	62.2	6741.22
Sweden	3.18	54.70	39.37	17.67	57.47	6072.88
Spain	6.38	60.51	44.35	25.59	68.17	10,085.97
Cluster B means	4.78	39.46	36.22	20.69	61.46	7602.12
Greece	1.33	41.01	25.64	8.88	54.3	14,135.79
Austria	3.63	23.50	35.49	19.37	54	14,032.85
Cyprus	1.22	54.54	40.51	25.14	69.2	16,474.96
Malta	7.55	58.52	56.51	26.43	72.6	18,249.74
Croatia	5.80	25.59	23.05	26.64	57.9	23,979.15
Cluster C means	3.91	40.63	36.24	21.29	61.60	17,374.50
Eu mean	3.57	33.85	31.26	16.71	55.25	7512.01

Source: author’s design using SPSS v.27.0.

in the Appendix A).

Cluster A includes countries such as Czechia, Germany, Luxembourg, Estonia, Bulgaria, Finland, Belgium, Lithuania, Hungary, Poland, Slovakia, Latvia, and Romania. Overall, this cluster demonstrates moderate to low digitalization levels, although internal variation is evident. For instance, Germany and Finland show higher rates of CRM and BI adoption, reflecting more advanced integration of digital tools, whereas Romania and Bulgaria remain at the lower end of the spectrum. The cluster’s average CRM usage (25.92%) and BI adoption (12.19%) fall below the EU mean, and its average Nights Spent at Tourist Accommodation Establishments (NSTA) of 3656.36 per 1000 inhabitants suggests that these economies are still in transition from traditional to fully digitalized tourism structures. Nevertheless, the average occupancy rate (48.51%) indicates that particular operational efficiencies persist independently of advanced technological adoption, likely driven by non-digital competitive factors such as affordability or natural tourism assets.

Cluster B encompasses the Netherlands, Portugal, Italy, Slovenia, Ireland, Denmark, France, Sweden, and Spain. These countries exhibit consistently high levels of technological integration, marked by CRM adoption rates averaging 36.22%, ERP systems at 39.46%, and BI tools at 20.69%, all exceeding the EU average. The relationship between digital maturity and tourism success is evident in this group: the average occupancy rate reaches 61.46%, while NSTA rises to 7602.12 nights per 1000 inhabitants. The alignment between technology adoption and tourism competitiveness in Cluster B highlights the strategic importance of CRM and ERP in driving service innovation, customer engagement, and destination attractiveness. CRM, in particular, emerges as a defining feature, enhancing personalized guest interactions and contributing to the sustained growth of tourism performance indicators.

Cluster C comprises Greece, Austria, Cyprus, Malta, and Croatia and represents a distinctive configuration where high tourism intensity coincides with strong, though not uniformly advanced, digital adoption. While CRM (36.24%) and BI (21.29%) are present at levels comparable to Cluster B, this group stands out primarily for its exceptional tourism activity. The average NSTA, at 17,374.50 nights per 1000 inhabitants, is far above the EU mean, accompanied by an average occupancy rate of 61.60%. The interaction between deeply rooted hospitality traditions and the strategic use of digital tools, especially CRM, appears to amplify performance outcomes. In Malta and Croatia, for example, CRM adoption exceeds 50%, underscoring a strong orientation toward customer experience management in highly seasonal tourism markets.

Taken together, these results offer robust empirical support for Hypothesis H2. Countries with higher levels of CRM adoption and integrated digital infrastructures consistently cluster in the higher-performing groups, indicating that digital transformation acts as a key differentiator of tourism competitiveness within the EU. The coexistence of digitally mature economies (Clusters B and C) and transitional ones (Cluster A) further highlights the uneven diffusion of digital innovation across Europe. Importantly, the findings suggest that CRM operates not merely as a technological instrument but as a strategic enabler of tourism performance, linking digital investment to tangible economic outcomes such as occupancy and length of stay.

5. Discussion

This research aims to deepen understanding of how digital technologies, especially artificial intelligence (AI), influence tourism performance across Europe by combining thorough data analysis with a critical review of existing scholarship. The study follows two main paths: it examines the link between different software solutions used in the hospitality industry and the performance of accommodations, and it classifies European countries based on their level of digitalization and tourism results. This approach not only identifies statistical relationships but also develops explanatory models that reveal the structural changes happening in modern tourism.

The statistical and clustering analyses confirmed through hierarchical methods support the first hypothesis (H1), which states that customer relationship management (CRM) software has the strongest link to tourism performance as measured by net occupancy rates and the number of overnight stays in tourist establishments. This result aligns with recent research highlighting CRM’s ability to gather, organize, and interpret detailed data on consumer preferences and behaviors [4,36]. Studies indicate that adopting such systems enables more personalized services and more efficient interactions with clients, which in turn increases satisfaction and loyalty. From this perspective, CRM is not only a tool for improving customer relationship management but also a direct factor in enhancing operational performance in tourism.

When it comes to AI-based technologies, the results show increasing relevance but a weaker direct link with tourism performance indicators compared to CRM. This outcome can be attributed to the emerging nature of many AI applications, which are still in the early stages of adoption across much of Europe. While the literature highlights numerous potential benefits of AI in automating, personalizing, and streamlining tourism processes [24], the actual pace and scope of implementation vary widely from country to country. For example, the studies by Elkhwesky et al. [1] and Doborjeh et al. [45] show that the applicability of AI heavily depends on local context, infrastructure availability, and the digital skills of the workforce.

The findings also align with the views of Liu et al. [43] and Song et al. [42], who observe that AI is more often used in forecasting, sentiment analysis, or automated management systems than in activities directly related to quantitative indicators of tourism performance. Over time, AI is expected to become more important as these technologies become more affordable and more deeply woven into the operational systems of tourism companies.

Validation of the second hypothesis (H2) provides further insight. Ward’s clustering method identified three groups of European countries based on their use of AI, ERP, CRM, and BI software, along with their respective performance in the tourism sector. This structure confirms the presence of regional typologies and supports the idea that the integration of digital technologies into tourism remains inconsistent and is influenced by each country’s specific economic, technological, and institutional conditions. The findings align with those of Agiropoulos et al. [11], who showed that economic and infrastructural factors play a key role in regional tourism success within the European Union [67].

The presence of clusters also aligns with the work of Streimikiene et al. [2] and Dolnicar [55], who emphasize that tourism performance and digital adoption do not follow a uniform pattern but instead mirror distinct regional models. Countries in Cluster B, characterized by high adoption of CRM and AI along with notable tourism performance, are generally mature economies with advanced digital infrastructure and innovation-friendly environments. In contrast, Cluster A groups countries in a transitional stage of digital adoption, with lower levels of technological integration but significant development potential.

This differentiation aligns with the conclusions of Rahmadian et al. [68] and Lv et al. [69], who assert that AI can enhance tourism sustainability, but only when a strong technological foundation and strategic vision are already in place. Without these elements, AI risks remaining just a superficial tool, used sporadically and lacking meaningful impact on performance. Similarly, the practical application of AI in ecotourism or forecasting platforms depends on organizational capacity and the support of public policies [70,71,72].

The differences between clusters also demonstrate the limited explanatory power of economic factors. Countries with higher GDP per capita, better infrastructure, and larger inflows of foreign direct investment tend to adopt technologies more quickly, aligning with the models developed by Antonakakis et al. [17] and Anagnostou et al. [3]. However, cultural and institutional differences cannot be ignored, as perceptions of AI often depend on factors like trust in technology, digital education levels, and openness to innovation [41,73].

The results of this research offer more than just empirical validation of the two hypotheses. They provide an interpretive framework that makes sense of the complex relationship between technological changes and tourism success. They demonstrate that the successful integration of AI into tourism does not happen automatically but relies on structural and contextual factors, with the benefits becoming most evident in countries that combine technological investments with a clear strategic vision.

This analysis also presents valuable insights for tailored policy-making at the EU level, providing guidance for fund allocation, training program development, and promoting public–private partnerships to speed up digital transformation. In this way, AI’s role in European tourism performance becomes not only a subject of research but also a practical mission focused on strengthening the sustainable competitiveness of destinations in the post-digital era.

5.1. Theoretical Implications

This study strengthens the theoretical foundations of digitalization in European tourism by providing a detailed view of how software solutions, especially artificial intelligence, impact the performance of accommodation providers. Unlike traditional methods that focus on a single technology, this research presents an interpretive model that combines four categories of digital tools, AI, ERP, CRM, and BI, and connects them to key tourism performance indicators. It not only confirms earlier findings about AI’s effectiveness in personalizing tourism services [7,10] but also creates an analytical framework that functions at a macro-regional level. The methodology, which merges artificial neural networks with cluster analysis, makes the exploration of nonlinear relationships and the identification of structural types among EU member states more robust. The theoretical contribution also broadens the literature by showing how CRM has become a differentiating factor in tourism competitiveness, providing empirical support for its strategic role in influencing occupancy patterns and increasing the length of stays.

5.2. Practical Implications

The empirical results provide clear guidance for tourism policymakers and industry stakeholders, highlighting future trends related to AI technologies. The confirmed impact of CRM on tourism performance indicates that investments in systems supporting personalized customer interactions produce tangible benefits, such as higher occupancy rates and increased customer loyalty. Tourism businesses can use these insights to strategically focus on digital development, adapting it to local conditions and levels of technological maturity. At the policy level, grouping European countries based on their digital profiles and tourism performance offers a valuable framework for regional and EU initiatives, helping ensure more efficient resource use and targeted transfer of best practices. Countries with weaker performance may benefit from targeted assistance to accelerate digital transformation, while more advanced regions could serve as models for those still transitioning. Overall, the findings emphasize that tourism digitalization progresses unevenly, requiring tailored strategies and sustainable partnerships between the public and private sectors.

5.3. Limitations and Further Research

Although the findings provide a compelling picture of the relationship between digitalization and tourism performance, certain methodological limitations require attention. The analysis relied on nationally aggregated data, which may conceal important regional differences vital for a deeper understanding of the phenomenon. Additionally, the lack of qualitative evidence on how software solutions are actually implemented within organizations limits the ability to capture varied impacts depending on technological sophistication. Another limitation comes from the static nature of the analysis, which identifies correlations at a single point in time without considering the dynamics of change or the effects of significant events such as health or economic crises.

Future research could address these issues by adding a longitudinal aspect, studying changes over time in both digital and tourism metrics, and incorporating regional or local data. Broadening the methodological approach toward explainable AI models would also enhance understanding of algorithmic decisions, which is essential for the sustainable and ethical use of these technologies. Conceptually, more exploration is needed to understand how AI influences not only economic outcomes but also the social and ecological sustainability of tourism, highlighting digitalization as a key element of responsible tourism in Europe.

6. Conclusions

This paper emphasizes how digital transformation, especially the adoption of AI-based solutions, changes tourism performance in Europe. It is not just a response to current market needs but also part of a broader modernization and adaptation strategy to current economic and social realities. The analysis shows that among the various digital tools used in the hospitality sector, CRM is key in achieving strong tourism results, including higher occupancy rates and more overnight stays. Meanwhile, AI technologies are increasingly important, signaling a shift in how guest relations and operational processes are managed.

The validation of the two hypotheses in this study provides empirical support for these conclusions while also offering a practical view of the digital diversity among EU member states. The clustering of countries by their level of digitalization and tourism performance revealed clear regional patterns, which are valuable for public policy development and creating strategies tailored to national contexts.

This research offers a reflection on the future of European tourism in the digital age. The results indicate that success in this field no longer relies solely on natural or cultural resources but increasingly on the ability to integrate new technologies intelligently and responsibly. In a landscape filled with uncertainty, digital innovation can catalyze resilience, competitiveness, and sustainability, so long as strategic decisions are based on data-driven insights, adaptability, and a deep understanding of the interaction between human experience and technological advancement.