Case Study: Unified Analytics Hub & Self-Service BI Platform

Retail

Unified Analytics Hub & Self-Service BI Platform

The Challenge: 23 Reporting Tools, No Single Source of Truth

A multi-brand retail group operating eight brands across 500+ stores had accumulated 23 separate reporting tools over a decade of organic growth and acquisition. Merchandising ran on one BI platform, finance on another, store operations on a third — and none agreed on definitions for core metrics like gross margin, comparable store sales, or sell-through rate. Weekly trading meetings regularly devolved into arguments about whose numbers were correct rather than decisions about what to do. The data engineering team received an average of 15 ad-hoc report requests per week, consuming 40% of their capacity. When requests were fulfilled, they typically took three to five days — too slow for the fast-moving merchandising decisions that determined profitability. Executive leadership had no real-time view of business performance: the weekly board pack was assembled manually by a team of two analysts working from Friday through Sunday each week.

Our Solution: Databricks Lakehouse with Certified Semantic Layer

Lydatum designed and delivered a unified analytics platform over seven months, built on a Databricks lakehouse with a Power BI semantic layer serving both executive and self-service needs:

• Unified Data Foundation: We migrated all 23 reporting data sources into a Databricks Delta Lake, building a medallion architecture (bronze, silver, gold) with dbt managing all transformation logic. Gold-layer certified datasets encoded agreed business definitions for every key metric, version-controlled and documented in a shared data catalog. The days of competing definitions were eliminated at the infrastructure layer.
• Executive Analytics Hub: We built a suite of Power BI executive dashboards refreshing hourly — P&L performance by brand and region, inventory health and sell-through rates, customer acquisition and retention trends, and supply chain fill rates. A mobile-optimized board pack was auto-generated every Friday morning, replacing the manual weekend process entirely.
• Self-Service Analytics Layer: We created 40+ pre-built Power BI report templates covering the most common analytical needs across merchandising, finance, and store operations. Business analysts were trained to build and publish their own reports from certified datasets without routing requests through engineering. A governance framework defined who could publish certified reports versus exploratory analyses, preventing data proliferation while enabling autonomy.
• Enablement Program: A structured data literacy program trained 200+ business users across six brands on Power BI self-service capabilities, data catalog usage, and metric definitions. An internal analytics community of practice was established to sustain momentum post-engagement.

The Impact: Data-Driven Decisions at the Speed of Retail

The platform went live across all eight brands simultaneously, with the executive dashboard suite active from day one. Within four weeks, the weekly manual board pack process had been fully retired.

Ad-hoc data engineering requests dropped by 80% within three months as business teams gained self-service capability. The data engineering team redirected freed capacity toward higher-value initiatives including predictive inventory optimization and customer lifetime value modeling. The merchandising team credited faster access to sell-through data with a 4-point improvement in full-price sell-through rate — worth an estimated $12M in annual margin.

Services Delivered: Dashboard & BI Development, Self-Service Analytics Enablement, Data Storytelling & Communication, Data Architecture & Modeling

Technologies Used: Databricks, Delta Lake, dbt, Power BI, Azure Data Factory, Great Expectations