Data spread across an ERP, a CRM, a project system and two dozen spreadsheets does not become useful by being copied into one place. It becomes useful when it is modelled - when a customer means the same thing in every table and a month closes on the same day everywhere.
We have built information layers integrating ERP data across nine countries, combined every acute hospital in a national health service onto one model, and merged databases held at different granularity into single fact tables that report daily. The methodology is dimensional and unfashionable, and it still outlasts the tooling.
What this includes
- Kimball methodology dimensional warehouse architecture
- Combining source systems held at different granularity into one coherent model
- Ingestion from ERP, CRM and project systems - Dynamics 365, Asta PowerProject and others
- User-driven file ingestion where the source is a person with a spreadsheet
- ETL and ELT engineering in PySpark, T-SQL, SSIS, Data Factory and Informatica
- Historic backload, so a newly onboarded source goes live with its full history
- Performance tuning of existing warehouses and query layers
- Row-level security and multi-tenant data models
- Documentation of models, lineage and business rules
What you end up with: One model the whole business reports from, that a new developer can read without a handover meeting.
Every engagement includes detailed documentation of our findings, and all source code and artefacts for any bespoke product - documented for your teams to maintain and understand. We maintain a strict level of confidentiality throughout.