How FLO Cut Lost Sales by 12% with AI-Driven Demand Forecasting

The Challenge

FLO is one of Europe's largest footwear retailers, operating more than 800 stores and a multi-brand e-commerce platform across 25 countries, with over 11,000 employees managing millions of SKUs each season. Balancing local fashion cycles, promotional calendars, and clearance assortments across both physical and digital channels meant planners needed precise, store-by-store inventory decisions made constantly, not occasionally.

Despite that scale, planners were still leaning on legacy, spreadsheet-driven processes that couldn't keep pace with real-time shifts in customer demand. Fragmented data across channels made it hard to know where stock was running low and where it was sitting excess, which translated directly into lost sales and inefficient markdown timing.

The Solution

FLO partnered with invent.ai to bring a unified, AI-driven planning platform into its operations, deployed on AWS. The system integrates sales data, web analytics, promotion schedules, and external signals such as local weather and events to produce real-time forecasts down to each SKU, store, and day, replacing static spreadsheets with continuously updated, granular demand signals.

• Real-time, granular demand forecasting at the SKU/store/day level, fed by sales data, web analytics, promotions, weather, and local events
• Automated transfer recommendations, flagging overstocked stores and routing inventory to locations running low
• Agentic AI for size optimization - clustering stores with similar selling patterns and fine-tuning case packs to reduce leftover odd sizes
• Markdown timing optimization as products near end of shelf life, protecting revenue instead of relying on blanket clearance
• Distribution network modeling to test changes to DCs, hubs, and store routing before making real-world changes

The Results

• 12% reduction in lost sales by keeping the right styles and sizes in stock at the right locations
• Static spreadsheet planning replaced with real-time, dynamic, store-by-store inventory insight
• Stronger sell-through and fewer leftover odd sizes through AI-driven case-pack optimization
• Improved operational efficiency and inventory positioning across 650+ stores and 15 distribution centers