How did Saint-Gobain Sekurit improve forecast accuracy with Flowlity?

Sekurit moved from macro-level sales estimates to SKU-level probabilistic forecasts powered by Flowlity, which is the structural shift behind the 15% improvement in forecast accuracy at SKU level across 10,000+ glass references. Before Flowlity, the team produced top-down forecasts based on aggregate market trends and budget assumptions, then disaggregated them to SKU level using historical splits. This worked for aggregate planning but generated systematic errors at SKU level, where each reference has its own demand profile, seasonality, and lead time. The new approach builds forecasts bottom-up at the SKU level using probabilistic AI that captures variability rather than averaging it away. The 15% accuracy gain translated directly into better replenishment decisions at distribution centers, smoother production planning at the three European plants, and a much more stable availability curve through the year, with the deep seasonal dips that previously disrupted service to garages and fitting centers largely eliminated. The team also developed comprehensive operational KPIs on top of the new platform, which connects directly to multi-echelon inventory optimization across the network.