General Merchandise Retailer (anonymous)

In retail, the gap between forecast and reality is measured in markdowns—and for a global general merchandise retailer operating 220 brick-and-mortar stores plus e-commerce channels across roughly 500,000 SKUs, that gap was costly. Fashion and seasonal merchandise require hyper-local demand intelligence: weather patterns, local demographics, pricing sensitivity, promotional timing, and assortment mix all shift demand at the store level. The retailer's incumbent Blue Yonder environment could not incorporate these external signals at scale. Worse, assortment, financial, and supply planning ran on disconnected systems, creating siloed decisions that compounded forecasting errors. The result was excess inventory, elevated markdown rates, and chronic stock imbalances that eroded gross margin.

The retailer replaced Blue Yonder with o9 Solutions' Enterprise Knowledge Graph, consolidating Forecasting, Allocation, and Replenishment onto a single AI/ML-based platform. The system ingests external demand drivers—weather, demographics, pricing, promotions, product assortment, and location—and synthesizes them into a unified forecast across all 500,000 SKUs and 220 stores simultaneously. Machine learning and predictive analytics power both regular and cross-dock store allocation workflows, with store capacity factored directly into inventory planning decisions. Omnichannel visibility spans brick-and-mortar and e-commerce in a single planning view. By migrating from a monolithic legacy architecture to o9's open-architecture platform, the retailer connected assortment, financial, and supply planning into one continuous decision loop for the first time.

The deployment delivered a step-change in gross margin performance, driven primarily by localized forecast precision that reduced the need for markdown intervention. Key outcomes included:

• Markdown reduction: Hyper-local demand signals cut excess inventory buildup for fashion and seasonal categories.
• In-stock improvement: Demand-driven replenishment raised service levels across stores and e-commerce.
• Planner productivity: Consolidating fragmented systems freed planners from manual reconciliation, shifting capacity toward higher-value decisions.
• Sustainability gains: Reduced overstock translated into lower excess inventory and more efficient capacity utilization across the network.

Collectively, these outcomes converted what had been a source of supply chain volatility into a competitive advantage in a high-SKU, high-seasonality environment.

• Unify planning before optimizing forecasts: Disconnected assortment, financial, and supply systems amplify forecast errors—consolidation onto a single platform is a prerequisite for AI to deliver margin impact.
• Hyper-local signals are non-negotiable for seasonal retail: Weather, demographics, and location-level data must feed the model directly; aggregate forecasts cannot substitute for store-level precision.
• Legacy replacement requires an open architecture: Monolithic systems like Blue Yonder resist integration with modern ML pipelines; open-architecture platforms enable faster iteration and omnichannel response.
• Store capacity belongs in inventory planning: Ignoring physical storage constraints upstream creates downstream replenishment failures regardless of forecast accuracy.