GXO Logistics

Managing inventory across 980 stores in 23 countries is an operational challenge that exposes the limits of manual planning systems. This retailer's existing tools lacked the sophistication to handle simultaneous constraints: shipment schedules, distribution sorter capacity, package prioritization, and store-level demand variability across dozens of markets. Planners compensated with labor-intensive workarounds, but the gap between what products were available on shelf and what customers wanted to buy translated directly into lost sales. In footwear retail — where size, style, and seasonal demand shift quickly — poor on-shelf availability is not a minor inefficiency. It is a measurable, recurring revenue leak.

The retailer partnered with invent.ai to deploy two customized AI-driven modules: Store Replenishment and Transfer Optimization. The replenishment module uses machine learning and predictive analytics to continuously calculate optimal stock quantities per SKU per store, accounting for real-time demand signals, lead times, and distribution constraints such as sorter capacity and shipment frequency. The transfer optimization module identifies imbalances across the store network — where excess stock in one location represents unmet demand in another — and generates transfer recommendations that minimize lost sales while controlling logistics costs. Both solutions were integrated into existing operational workflows and fully deployed within six months, replacing the manual processes that had previously required significant planner hours to maintain.

The combined deployment delivered measurable improvements across the retailer's key inventory health indicators within the deployment period:

• Substantial additional sales revenue generated through improved product availability and optimized inter-store transfers
• A significant reduction in lost sales, directly attributable to fewer stockouts at the store level
• 8.8% improvement in on-shelf availability, ensuring the right sizes and styles were present where demand existed
• 4% increase in additional sales representative unit ratio, reflecting the revenue captured through smarter transfer decisions

Beyond the financials, automated replenishment and transfer recommendations reduced the manual burden on planning teams, reallocating workforce capacity toward higher-judgment tasks.

• Addressing replenishment and inter-store transfers as a unified system — not separately — compounds impact; availability gains reduce lost sales while transfer optimization captures revenue from inventory already in the network.
• At scale (980+ stores, 23 countries), operational constraints like sorter capacity and shipment windows must be embedded directly into the AI model, not treated as post-hoc filters.
• A six-month deployment timeline is achievable when the vendor builds to the retailer's specific constraints rather than fitting the retailer to an off-the-shelf model.
• On-shelf availability improvements create a feedback loop: better availability data improves demand signal quality, which further refines replenishment accuracy over time.