Increase in forecast accuracy
Contribution to company revenue
With consumers expecting a high degree of variety in the products hitting the shelves, product aisles are no longer lean. This has made it increasingly difficult for CPG companies to manage their supply chains.
Our client operates in South and East African markets, supported by a network of production centres, logistics teams, regional sales teams, and localized sales representatives spread across strategic locations in these regions.
Often, teams struggle in developing a robust chain that delivers what the consumer seeks. However, digital adopters have made large strides by adopting advanced forecasting solutions to better manage their distribution networks.
There was significant opportunity in driving more value out of the existing network by being able to anticipate future demands. This would help integrate sales and production teams, ensuring that key products would hit the shelves as expected.
The first two problems “What to produce” and “How much to produce?” were solved by forecasting the revenue and units for SKUs across Categories for the next quarter. The third problem focused on aligning sales and production teams by letting them apply overrides within acceptable limits.
AWS Deep AR+ model (with RTS) is used as the forecasting engine (Store-SKU-Day level forecast)
Mathematical equation incorporating SOH, dump, planogram (Visual Display), seasonal availability & other category inputs are included in the model
Apache Airflow schedular is designed to orchestrate forecasting, indenting activity
The solution was first piloted across a country to validate the accuracy of our forecasts and to scout for process improvements. The solution was then scaled up to various countries in a phased manner across East and South African markets.
12% increase in forecast accuracy (baseline of 65%) for key Categories contributing to 98% of revenue.
The automated engine shortened the quarterly planning cycle, and also made it easier for internal teams to set out accurate production targets.