At a glance
- Client
- A consumer goods distributorName withheld for confidentiality
- Industry
- Consumer Goods & Distribution
- Service
- Smart Business Operations
- Scope
- Demand forecasting models and replenishment automation
The challenge
Purchasing relied on last year's sales and gut feel. Fast-moving products ran out during promotions and festivals, while slow movers filled warehouse space and tied up cash.
Each stockout meant lost sales and unhappy retailers; each overstock meant discounts and write-offs.
Our approach
Gather the demand signals
We combined sales history, promotions, seasonality and festival calendars by product and location.
Forecast at SKU level
Machine learning models predict demand for every product in every warehouse.
Automate replenishment
Forecasts feed reorder points and suggested orders in the ERP.
Review exceptions
Planners approve or adjust only the orders that look unusual.
The solution
Replenishment is now driven by forward-looking forecasts instead of past averages. Suggested orders are generated automatically, and planners focus on new products, promotions and exceptions.
Stock levels follow real demand more closely, freeing cash and warehouse space while keeping shelves filled.
Results
- SKU-level forecasts for each warehouse
- Festivals and promotions built into predictions
- Automated suggested orders in the ERP
- Planners focus on exceptions instead of spreadsheets
Technology used
- Python
- LightGBM
- Time-series forecasting
- ERP integration
- Power BI
Facing a similar challenge?
Tell us what you are working on. We will walk you through how we would approach it, with no obligation.