Retail
Enterprise Client (Anonymized) — Multi-Format Retail
Cutting stockouts on top SKUs by a quarter across 180 stores
26% reduction in stockouts on top-selling SKUs
Deployment · 10 weeks
The situation
A 180-store retail chain was running replenishment on a forecasting model that hadn't kept pace with how the business had grown — seasonal spikes were consistently missed, top-selling SKUs stocked out at a rate that was costing measurable revenue, and the planning team spent most of their week manually overriding system-generated orders they didn't trust.
Leadership wanted better forecasts. What they needed was a system planners would actually use instead of routinely overriding.
The baseline
- Forecast accuracy (top 500 SKUs): 61% (MAPE ~39%)
- Stockout rate on top-selling SKUs: 14%
- Planner time spent on manual overrides: roughly 60% of the planning week
- Replenishment cycle: weekly, store-level, largely rules-based
The deployment
Two engineers embedded with the merchandising and planning team for ten weeks. We built a forecasting model trained on store-level sales, promotions, and local seasonality, integrated directly with the existing ERP and inventory system — planners kept their existing tools, the forecast feeding into them just got dramatically better.
Critically, we built the recommendation layer to show its reasoning: which factors moved a forecast up or down for a given store and SKU, so planners could trust an override decision instead of defaulting to manual planning out of habit.
The outcome
- Forecast accuracy: 61% → 84% (MAPE down to 16%)
- Stockouts on top-selling SKUs: down 26%
- Inventory carrying cost: down 12%
- Planner manual-override time: down 70%, freeing the team to focus on promotional and new-product planning instead of routine replenishment
What the client owns now
- The forecasting pipeline and its store/SKU-level feature set
- The replenishment recommendation logic and its explainability layer
- The planner-facing dashboard, integrated into existing planning tools
- A runbook for retraining ahead of major seasonal or promotional shifts
Client identity, sector specifics, and select figures in this case study have been altered or composited to protect client confidentiality, standard practice for published engagement work. The deployment, methodology, and results described reflect a real CopperPin engagement.