A Multidimensional Analysis of User Experience and AI Personalization in Indonesian University Websites

Technical Summary: A Multidimensional Analysis of User Experience and AI Personalization in Indonesian University Websites

Problem Statement
The rapid digital transformation of higher education, accelerated by the COVID-19 pandemic, has established university websites as critical interfaces for student engagement, institutional branding, and international recruitment. However, existing literature on university website evaluation remains fragmented, often focusing narrowly on technical usability or general satisfaction while neglecting the integration of content strategy, international perception, and emerging Artificial Intelligence (AI) personalization features. This gap is particularly acute in the Indonesian context, where distance education institutions face systemic challenges regarding digital infrastructure, quality assurance, and the need to optimize web-based communication for both domestic and international audiences under the "Merdeka Belajar" philosophy. There is a lack of empirically grounded, multidimensional frameworks that evaluate how these distinct domains interact to shape user experience (UX) in AI-enabled educational platforms.

Methodology
This study employed a cross-sectional mixed-methods convergent design to evaluate user perceptions across four Indonesian higher education institutions: Universitas Terbuka, Universitas Negeri Surabaya (UNESA), Universitas Pradita, and the Indonesia Cyber Education Institute (ICEI).

• Participants: The final sample consisted of 75 respondents, comprising Indonesian citizens (n=44, 58.7%) and international users (n=31, 41.3%) from eight countries (Philippines, Japan, Myanmar, Sri Lanka, Pakistan, Mexico, Timor-Leste, and Turkey).
• Instrument: Data was collected via a structured online questionnaire hosted on Google Forms, featuring 62 rating items organized into five constructs and four open-ended qualitative questions. The rating items utilized a seven-point bipolar semantic differential scale (1 = negative attributes to 7 = positive attributes).
• Constructs: The evaluation covered four primary domains: Content Strategy, Usability, International Perception, and AI Personalization, alongside an Overall Satisfaction metric.
• Analysis: Quantitative analysis involved descriptive statistics, Cronbach's alpha for internal consistency, Pearson correlation, independent samples t-tests, one-way ANOVA, and multiple linear regression using Python libraries (scipy, statsmodels, pandas). Qualitative data underwent thematic analysis to identify priority improvement areas.

Key Results
The study yielded high internal consistency reliability across all constructs (Cronbach's α ranging from 0.929 to 0.982).

• Descriptive Statistics: All four evaluation domains received moderate-to-high satisfaction scores (means > 5.0 on a 7-point scale). AI Personalization achieved the highest domain mean (M=5.26), followed closely by Content Strategy (M=5.20) and International Perception (M=5.20), with Usability slightly lower (M=5.13).
• Regression Analysis: A multiple regression model revealed that the four domains collectively explain 87.6% of the variance in Overall Satisfaction (R² = 0.876).
  • AI Personalization emerged as the strongest unique predictor (β = 0.512, p < .001).
  • Usability was the second significant predictor (β = 0.189, p = .007).
  • Content Strategy and International Perception did not reach statistical significance as unique predictors when controlling for the other domains, though they remained highly correlated with satisfaction.
• Sub-dimensional Findings: Within Content Strategy, "Accommodation and living costs" received the lowest rating (M=4.36), indicating a critical information gap for international students.
• Comparative Analyses: No statistically significant differences were found between domestic and international respondents across any domain, suggesting cross-culturally consistent perceptions. However, a significant gender difference was observed in AI Personalization perception, with female respondents rating it higher than males. Universitas Terbuka achieved the highest overall satisfaction scores (M=6.07) compared to the other institutions.
• Qualitative Insights: Thematic analysis highlighted that users value ease of use and clear information presentation. Key improvement requests included enhanced server performance (24.2% of enhancement comments), better navigation, improved search functionality, and more robust personalization features (e.g., chatbots, adaptive recommendations). International users specifically requested multilingual accessibility and professional translation over automated tools.

Key Contributions

1. Multidimensional Framework: The study introduces a comprehensive evaluation framework that simultaneously assesses content strategy, usability, international perception, and AI personalization, moving beyond traditional, siloed usability metrics.
2. Predictive Validity of AI: The research provides empirical evidence that AI personalization is the dominant driver of user satisfaction in the context of Indonesian university websites, surpassing traditional content and usability factors in predictive power.
3. Cross-Cultural Consistency: The findings challenge assumptions of significant cross-cultural variation in website preferences, demonstrating that well-designed platforms can achieve comparable user experiences across diverse national backgrounds.
4. Contextual Specificity: The study addresses a gap in literature regarding Indonesian distance education, offering specific data on the performance of both public and private institutions in a developing digital infrastructure context.

Significance and Claims
The paper claims that its findings represent a paradigm shift in how digital educational environments should be designed and evaluated. By identifying AI personalization as the strongest predictor of satisfaction, the study suggests that the Technology Acceptance Model (TAM) may require updating to explicitly incorporate AI personalization as a distinct, high-impact predictor construct.

Practically, the results offer actionable guidance for institutional administrators and web developers. The authors argue that institutions should prioritize investments in chatbot development, personalized content recommendation systems, and adaptive search functionality to enhance user engagement. Furthermore, the study highlights the urgent need to address specific content gaps, such as accommodation and living cost information, and infrastructure limitations like server performance, to support international student recruitment and retention. The research concludes that systematic attention to website quality, particularly the integration of effective and trustworthy AI personalization, is a strategic imperative for institutions seeking to enhance their international reach and deliver equitable digital learning experiences.