Exploring the Role of Interaction Data to Empower End-User Decision-Making in UI Personalization

Paper Title

Exploring the Role of Interaction Data to Empower End-User Decision-Making in UI Personalization

Publication Info

• Topic area: User interface personalization through interaction data and user-driven decision-making.
• Keywords: UI personalization, interaction data, user-driven personalization, system support, visual suggestions, data transparency, cost-benefit analysis, user agency, privacy, mixed-initiative systems.

Background and Problem

• Problem / challenge: Despite the potential of user-driven UI personalization to enhance accessibility, efficiency, and user satisfaction, many users struggle to identify meaningful personalization opportunities due to limited support, effort required, and lack of perceived benefits.
• Significance: Addressing these challenges can empower users to take control of their digital interfaces, improve usability, and foster long-term engagement with personalized systems.
• Motivation and related work: Prior work has explored system-driven and user-driven personalization, but user-driven approaches often lack tools to help users identify and evaluate personalization opportunities. Interaction data has been used in system-driven personalization but remains underexplored as a resource for empowering user-driven personalization.

Solution

• Proposed approach: A reflexive personalization framework that leverages interaction data to support users in identifying and implementing meaningful UI personalization opportunities.
• Novelty:
  1. Investigates how interaction data can empower users to make informed personalization decisions.
  2. Explores the role of system-initiated visual suggestions in supporting user-driven personalization.
  3. Examines privacy and trust concerns related to interaction data usage in personalization.
  4. Proposes design considerations for integrating interaction data into personalization systems.
• Procedure and key techniques:
  • Conducted a semi-structured interview study with 12 participants using 42 experimental vignettes as design probes.
  • Explored four scenarios (A: data-only dashboards, B: textual suggestions, C: visual suggestions, D: social suggestions) to examine user responses to different levels of system support.
  • Analyzed participant feedback to identify themes related to data engagement, personalization benefits, and system expectations.

Results

• Concrete findings:
  • Most participants could independently identify personalization opportunities using interaction data, particularly click and scroll heatmaps.
  • System-initiated visual suggestions were preferred, as they reduced effort and increased confidence in personalization decisions.
  • Cost-benefit metrics (e.g., time savings) were valued for motivating personalization and assessing its benefits.
  • Privacy concerns were mitigated when users had control over data collection and could review how their data informed suggestions.
• Advantage over baselines:
  • Visual suggestions and cost-benefit metrics enhanced user engagement compared to data-only dashboards or textual suggestions.
  • Access to interaction data increased transparency, trust, and user agency compared to traditional system-driven personalization.
• Experiments / evaluation:
  • Participants engaged with vignettes representing four personalization scenarios, reflecting on their potential benefits, challenges, and expectations.
  • Data collected through interviews was analyzed thematically to identify user needs and preferences.
• Limitations and future work:
  • Synthetic data and fictional personas may have limited the depth of discussion about real-world personalization and privacy concerns.
  • Future work should conduct in-the-wild studies with real interaction data and explore broader user demographics and system configurations.

Summary

This study explores how interaction data can empower users to identify and implement UI personalization opportunities. Participants preferred system-initiated visual suggestions supported by cost-benefit metrics, which enhanced transparency, trust, and user agency. Privacy concerns were mitigated when users had control over data collection and usage. The findings highlight the potential of interaction data to foster proactive personalization and propose design considerations for future systems. Future research should validate these findings in real-world settings and expand the scope to diverse user groups and system designs.