Snowflake warehouse and operational reporting for a mid-market retailer

The problem

The retailer’s data lived in a half-dozen systems: POS, e-commerce, inventory, marketing, and finance. Reports got cobbled together in spreadsheets every week, numbers rarely tied out across teams, and operators made decisions on data that was anywhere from a day to a week stale. Leadership wanted one place to see what was actually happening across locations and channels, every morning.

The approach

We designed and built a Snowflake-based data warehouse with dbt-modeled layers and operational dashboards built directly on top. The system pulls from every operational source on a daily cadence, normalizes and reconciles across locations and channels, and exposes modeled tables to a BI surface operators trust because the numbers tie out, the latency is predictable, and the lineage is visible.

Architecture

[ Architecture sketch placeholder, replace with diagram when ready. ]

• Ingestion: Per-source connectors for POS, e-commerce, inventory, marketing, and finance systems. Daily incremental loads.
• Warehouse: Snowflake. Raw, staging, and modeled layers with full history retention.
• Transformation: dbt for all modeling, with tests on every business-critical metric.
• BI surface: Operational dashboards for store ops, e-commerce, merchandising, and finance, each scoped to the decisions that team actually owns.
• Governance: Documented lineage, tested metrics, role-based access. The data team owns the layer, operators consume it.

Results

• Daily operational reporting across all locations and channels, with numbers that tie out across teams.
• Hours of weekly manual report assembly eliminated.
• A trusted analytical foundation downstream teams (forecasting, customer analytics) can build on without standing up their own data plumbing.

Stack

Snowflake, dbt, daily orchestration, BI layer (Looker / Sigma / custom depending on team), per-source connectors.