Data exists in abundance in most pharmaceutical companies; the scarcity is in integrated, actionable intelligence. At SterisCare, sales data lived in a CRM, production data in an ERP, quality data in a LIMS, and field force data in a standalone mobile app. Each system had its own reporting logic, different data definitions, and separate access controls. Leaders spent more time reconciling reports than acting on them.

The project brief was clear: build a platform that makes the right data available to the right person, in the right format, without requiring IT intervention for every report request.

Our approach began with a data audit—a structured exercise that inventoried every key business metric, its source system, update frequency, data quality rating, and the decisions it was meant to support. This produced a 120-row data catalogue that became the foundation for all subsequent work.

The architecture we designed was a cloud-based data warehouse (Snowflake) with source-specific ELT connectors—both custom API connectors and native integrations where available. A lightweight orchestration layer (Apache Airflow) managed scheduling, error alerting, and lineage tracking. The semantic layer—built in dbt—ensured that every metric was calculated consistently regardless of which dashboard surfaced it.

Dashboard development was done in two phases. Phase one covered the four highest-priority use cases identified during discovery: sales performance (territory, product, customer), inventory and supply, manufacturing output, and quality trends. Phase two extended coverage to marketing ROI, order-to-cash, and HR metrics.

User adoption was addressed through a structured change management programme: early involvement of key users in requirements definition, a super-user network of 12 champions across departments, in-person training workshops, and a self-service help portal with recorded tutorials.

The platform launched on schedule, with 95 active users logging in within the first 90 days—a number that represented nearly every manager and above in the company. The monthly consolidated report, which previously took two analysts 40 person-hours to assemble, is now auto-generated in under four hours with automated anomaly flags.