Project Brief
Manufacturer Data Integration Platform
Quarterly distributor files in mixed formats were prepared by hand, transformed inside Power BI, and matched with partial Excel validation — creating a bottleneck around one technical resource.
Context
What made the work consequential.
Distributor sales arrived as Excel, CSV, TXT, and PDF; cleansing and matching depended on manual prep and Power BI-as-ETL; CRM reinjection and ERP extraction lacked a maintainable, traceable pipeline suited to a Microsoft-stack SME.
Key decisions and interventions
Manual close work became an explainable pipeline.
Designed a landing zone with Bronze/Silver/Gold layers so Power BI consumes Gold only
Specified multi-level matching with an exception queue instead of silent Excel validation
Scoped PDF/OCR assist for confirmed layouts while keeping deterministic pipelines as source of truth
Planned Fidelio ERP extraction, Zoho CRM write-back with a posting ledger, and parallel run until scorecard pass
Artifacts and capability
The operating model gained a governed data spine.
- Target architecture and medallion pipeline design
- Matching and exception-handling operating model
- ERP extract and CRM write-back integration plan
- Acceptance scorecard and knowledge-transfer approach
Current implementation evidence
What the available evidence supports.
The designed engagement is expected to cut the ~2-week manual close toward days after design freeze, raise explainable match coverage, keep reconciliation within planning tolerances, and demote Power BI to consumption-only. Business outcomes have not yet been independently verified.