A leading grocery supermarket chain wanted to reduce its inventory wastage and sales opportunity loss – the two main outcomes of inaccurate demand forecasting. It sought accurate predictions to optimize the inventory of its highly perishable items. Since sales of such items are highly dependent on price fluctuations, weather conditions, and seasonal demand patterns, among other factors, the company knew it needed an accurate prediction model to minimize its losses. With this goal in mind, it approached Netscribes to gain timely and accurate predictions to optimize its replenishment cycles across outlets.
Business challenge
Knowing exactly how much inventory to carry is paramount to grocery retail success. Overstocking can lead to loses due to wastage, while understocking translates into missed opportunities. Given that perishable goods have quick replenishment cycles, product indentation and allocation also needs to be quick. To overcome these challenges, a leading grocery supermarket chain turned to data analytics to predict the demand of perishable items and plan their inventory accordingly.
Approach
- Netscribes’ team of analysts synthesized the supermarket’s historical data on past sales, weather conditions, holiday sales, price changes, etc.
- The data was then treated to neutralize the effects of price fluctuations and seasonality in order to make it analyzable.
- After trialing various predictive models, a combination of time and price elasticity-based models were selected and applied to ensure the estimate garnered was as precise as possible.
Results Delivered
- A recommended inventory quantity was sent to each store by subtracting the forecasted sales from their stock in hand.
- The performance of these prediction models was tracked over a period of 12 months to ensure maximum accuracy.
Read more: Shrinkflation and its impact on consumer sentiment
Benefits
By implementing the recommended inventory planning model, the supermarket chain achieved:
- A highly accurate sales forecasting model with an error rate (MAPE) of only 4% against an industry-accepted rate of 15%
- A significant reduction in waste, contributing to an annual savings of USD 1.8 million
- Precise and reliable forecasts every seven days without any human intervention
