Power BI Analytics Platform — L'Oréal

Context & Challenge

During a consulting mission at L'Oréal, the Power BI reporting environment had accumulated years of technical debt. Critical DAX measures were slow — some taking several seconds to calculate on standard hardware — due to complex nested logic and redundant column calculations. The data model had grown organically into an over-normalized web of tables with ambiguous relationships rather than a clean star schema, making it difficult to add new metrics without unintended side effects. Manual data entry processes fed key KPI source tables across strategic reporting domains, regularly introducing errors that required time-consuming corrections and undermined trust in the reported figures.

Approach

The engagement started with a performance audit: profiling DAX measures using Performance Analyzer to identify the slowest calculations, mapping the existing data model to find redundant relationships and misplaced logic, and reviewing the data entry workflows that fed the reports. I rewrote critical DAX measures following best practices — moving from calculated columns to measures where appropriate, introducing variables to avoid redundant computation, and eliminating deeply nested CALCULATE chains by restructuring underlying model logic. The data model was redesigned as a proper star schema: one fact table per business process, dimension tables for shared attributes (dates, entities, cost centres), with clean, documented relationships and no circular dependencies. Power Automate flows replaced the manual data entry steps for KPI source tables, automatically populating records from standardized input forms and applying validation rules before data reached the Power BI model.

Results

Report load times dropped significantly after the DAX optimization and model restructuring, making the dashboards responsive for daily use in management meetings. KPI reliability improved as Power Automate flows eliminated the class of errors caused by manual copy-paste entry, reducing data correction requests from the business. The star-schema redesign gave the analytics team a stable, documented data model foundation — new KPIs could be added without risking regression in existing measures. The platform now serves as the single source of truth for real-time KPI monitoring across L'Oréal's strategic reporting scope.

This project was delivered by Klomèna YEO, freelance Analytics Engineer specialised in Finance & HR data. Available for similar engagements in France and internationally.