Demand forecasting that cut overstock by 34%
Results at a glance
34%
reduction in overstock costs
28%
fewer stockout events
$62M
annual savings in Year 1
The Challenge
A multinational retailer was losing $180M annually to overstock and stockouts across 4,200 SKUs in 14 markets. Their legacy forecasting model ran weekly and couldn't account for local demand signals.
What We Built
We built a real-time demand forecasting system using gradient-boosted trees trained on 6 years of sales history, enriched with weather, local events, and social trend signals. The model runs hourly and feeds directly into their replenishment system.
The Results
Project details
Client
Global Retail Group
Industry
Retail & E-commerce
Service
Custom ML Solutions
Duration
14 weeks
Ready to see similar results?
Let's talk about your data challenge
Every engagement starts with a free 30-minute discovery call. No pitch — just an honest conversation about your problem and whether we can help.