We rebuilt Costa’s data architecture to drive growth
Costa now has full control over their data, inlcuding real-time insights, reliable reporting, and a scalable foundation for growth.
The challenge: Bringing a fragmented data infrastructure in-house
Costa Coffee decided to bring their data infrastructure in-house. This strategic move aimed at improving control, flexibility, and long-term scalability. Up until then, much of their data architecture had relied on an external vendor, but as the company grew, so did the need for an internal solution they could fully own and evolve.
The transition came with its share of complexity. Data was distributed across two tenants, and documentation was limited. With the external partner stepping out of the picture, Costa needed support to rebuild the entire data warehouse from the ground up – fast, reliable, and future-proof.
The solution: Migrating and modernising Costa’s data architecture
We stepped in with strong domain expertise in data warehousing, ETL architecture, data QA, and data business analysis.
One of the first major steps was migrating data from an entire tenant into Costa’s new internal data warehouse. Alongside that, we designed and implemented new ETL pipelines, introduced clear documentation and data standards, and fixed inherited technical and logical issues — including the NPS reporting logic, which had critical flaws.
As part of our long-term support, we:
- developed and maintained multiple new data sources and pipelines
- corrected data quality issues left behind by the previous setup
- advised on architecture, often stepping in for the missing data architect
- restored key reporting functionality mid-migration, keeping the engine running while rebuilding it
The result: Real-time insights and full control over data
Costa successfully migrated one tenant's data to their internal data warehouse, with the next migration already underway. Several new data sources and reporting projects, previously not in place, were developed and launched. The handover from the previous vendor was fully completed, replaced by an internal, scalable solution.
What started as a vendor transition evolved into a long-term data enablement initiative.
The tech stack
- Python
- Databricks
- Azure Services: Data Factory, Synapse Analytics, Storage Accounts
- Apache Spark
- Azure DevOps used for workflow management (Stories in Boards can be linked to all other Azure system parts)
