Abstract
Understanding how users navigate complex digital platforms is essential for improving conversion, usability, and overall experience. Raw clickstream data generated by web platforms is highly granular, sparse, and noisy, which makes direct URL-level analysis unreliable and computationally unstable. In addition, traditional analytics approaches rely on aggregate metrics or predefined funnels that fail to capture sequential dependencies and heterogeneous user intent. This disclosure describes a method for identifying navigation inefficiencies, behavioral drop-off patterns, and experience deficiencies by combining probabilistic modeling of user navigation sequences with AI-driven synthetic personas. The method models user sessions as discrete-time Markov processes over a reduced semantic state space, identifies latent behavioral segments using unsupervised clustering, translates those segments into evidence-based personas through a structured marketing-research interpretation layer, and operationalizes the resulting personas as large language model agents (LLM) that analyze live web pages from empirically grounded behavioral perspectives.. The disclosed system enables detection of funnel bottlenecks, prediction of likely next navigation states, and scalable persona-based qualitative evaluation of web experiences.
Creative Commons License
This work is licensed under a Creative Commons Attribution-Share Alike 4.0 License.
Recommended Citation
INC, HP, "IDENTIFY USER NAVIGATION PATTERNS AND DIGITAL EXPERIENCE DEFICIENCIES USING PROBABILISTIC SEQUENCE MODELING AND LLM-DRIVEN SYNTHETIC PERSONAS", Technical Disclosure Commons, (August 18, 2026)
https://www.tdcommons.org/dpubs_series/11413