Abstract
The paper presents the modernization of a shopping mall marketing data analytics system originally implemented as a monolithic web application on .NET Framework with a single MS SQL Server database and batch integrations via file exchange. The audit identified limitations of the legacy solution a delay of more than 24 hours before data appeared in reports, performance degradation during peak periods, and high coupling of modules, where failures of individual components led to downtime of the entire platform. To reduce risks, an iterative Strangler Fig strategy was adopted with parallel operation of the monolith and new services. The target architecture is built on ASP.NET Core microservices with domain-based data isolation, the use of PostgreSQL for transactional data and Redis for hot metrics, as well as an event pipeline on Apache Kafka for near-real-time data processing and aggregation, followed by publishing indicators to BI dashboards. Implementation results demonstrate a reduction in data latency to seconds, increased resilience and release manageability, reduced MTTR to a few minutes, lower operating costs, and the economic feasibility of the project with payback within a 12-24 month horizon, including expansion of analytics through social media monitoring.