Grocery retailers manage tens of thousands of SKUs across 100s of locations. Owing to this sheer scale and complexity, out of stock situations have become alarmingly common in the grocery retail sector. In fact, some research suggests that the revenue loss due to going out of stock can be as much as $47 billion annually.
Many grocery retailers react to out of stock situations by increasing replenishment related parameters. However, these become prohibitively expensive and can lead to significant wastage in some categories.
What leads to stock-outs and wastage?
Demand is influenced by multiple drivers such as price, promotional activities etc., as well as external factors such as weather and macro-economic variables. Since this sensitivity is neglected in conventional processes such as the replenishment model, companies are at the risk of running into stock-out or excess stock situations.
The viable solution to this is through a proactive supply chain synchronization process. However, this depends on highly accurate forecasting done at outlet-SKU level on daily/weekly basis.
What are the limitations in demand planning with conventional forecasting methods?
Most times, demand forecasts are based on trend and seasonality observed over the years. This simplistic understanding of demand can lead to erroneous estimates. Also, any adjustments based on subjectivity and intuition fails the test of scale.
What is FORECAST Squared?
FORECAST Squared is a forecasting system that can build 1000s of highly accurate forecasts within a few minutes! Moreover, it does not assume any forecasting or data science background for users. Developing 1000s of highly accurate forecasts can be a matter of just a few clicks.
FORECAST Squared can generate monthly, weekly or daily forecasts for long time horizons. It also continuously updates forecasts based on the latest data.
How does FORECAST Squared work?
It considers many factors that can influence forecast. For an example, it considers factors such as advertising spend, price and promotions, weather and inflation. Next, it automatically considers many derivatives of these factors, which is called feature engineering in machine learning. Then the algorithm sweeps through 1000s of model classes and distinct parameter values and selects the best. This entire custom model building process is carried out without any human intervention.
- Highly accurate daily forecasts for perishable categories can help grocery retailers reduce wastage significantly whilst maintaining the correct level of inventory.
- For FMCG products, weekly forecasts can lead to significant improvements in supply chain synchronization