From Personalization to Conversion: Data-Driven Approach to Optimize the User Journey

Digital commerce is increasingly a competition for attention and relevance: every click, conversation or abandoned cart is a signal with economic consequences. This thesis investigates how organisations can turn those signals into repeatable business value by combining human-centred interaction design, data-driven personalisation and modern Generative AI architectures. Rather than treating models or interfaces in isolation, the work focuses on the operational “glue” — the processes, data pipelines, experiment designs and governance routines — that are required to move from prototype to measurable production impact. Using a mixed-methods approach, the research synthesises HCI and UX theory with industry benchmarks and translates these insights into a production-oriented pilot specification for a multi-channel conversational orchestration layer. Key research questions address (1) the impact of personalised and conversational interventions on core UX and business metrics , (2) architectural trade-offs for ML and GenAI solutions), (3) operational practices that enable safe and scalable deployment, and (4) the transferability of the approach across industries. The main contributions are threefold: (a) a coherent operational framework that links cognitive design principles, data architecture and experimentation to conversion KPIs; (b) a replicable pilot blueprint — including data flows, KPI definitions, A/B testing protocol and governance artifacts — demonstrated under realistic, benchmarked assumptions; and (c) pragmatic managerial guidelines and a roadmap for scaling conversational AI while controlling risk (privacy, accuracy, brand safety). By bridging theory and practice, the thesis aims to provide both scholarly insight into the integration challenges of UX, ML and GenAI, and actionable guidance for managers seeking to convert clicks and conversations into measurable revenue uplift.