The Opportunity
- An automotive company’s legacy parts forecasting system suffered from age, scalability issues, and expensive licensing fees to the vendor.
- During a short planning cycle at the end of each month, the demand planners needed to produce accurate forecasts for 225,000 parts in 17 distribution centers.
- The legacy system needed to be replaced.
Our Approach
We deployed a:
- highly parallel
- large-scale time series forecasting system
to produce “best pick” forecasts for 2.9 million part/location combinations.
The Impact
- Nearly three million forecasts are produced in thirty-five minutes. This allows planners to spend valuable time focusing on key parts that need a human touch.
- The company experienced a 33% improvement in accuracy in the first months after cutover.
- The IT department declared the project the most successful in years.