IA’s AI-guided forecasting model “ADA” decreased forecasting error for a large specialty retailer from 40% to 10% in 3 weeks time.
Forecasting in the best of times is challenging, with COVID it’s even further broken. Forecasting Accuracy is imperative in making business decisions that drive margin but it has declined rapidly in the post COVID environment. The retail industry has seen a remarkable change in 2020 like a shift in customer preferences, changes in customer demographics, etc. that make traditional forecasting methodologies useless. It’s pretty clear that retailers will need to evaluate their capabilities when it comes to forecasting. History is no longer a determining factor for how the future will play out. This is where AI, ML, and Predictive Analytics become imperative in forecasting and driving your merchandise strategies. The post-COVID world looks to be tough to navigate without the advanced analytical abilities that come with solutions that leverage AI and machine learning technologies.
The client is a leading US-based Speciality retailer. The client was grappling with changing shopper preferences and habits, and it became more challenging than ever to get the forecasts right with traditional methods and historical data. Some major challenges were-
- Forecasting methodology relied heavily on recent ROS and applying the LY sales trend.
- Large shift in categories selling
- Forecasting error resulted in allocating inventory to the wrong locations with half of the chain turning 50% too fast and the other half carrying twice as much inventory as needed.
- Demand variations across channels and categories.
- Current models of forecasting lack COVID adjusted elasticities and are inadequate in handling volatile business nuances.
It was important to employ AI-Powered forecasting models to account for the ever-changing environment. In other words, ADA was exactly what the client needed to get their forecasts right. ADA is an Impact Analytics AI Guided forecasting Model that uses an ensemble of AI algorithms and Machine Learning systems. ADA generated actionable and reliable forecasts for the client by-
- Incorporating internal and external factors like trends, seasonality, events, weather, marketing spends, panel data, and store closures.
- Imputed COVID-19 risk based on multiple external factors.
- Using an ensemble of complex AI-powered algorithms to design models
- Ensuring the automated forecast model works equally well in volatile circumstances- Rubber duck forecast adjustment.
- Taking recency much more into account that helped to continually analyze and adapt to the latest market conditions.
ADA enabled clients to achieve forecasting accuracy and help them gain real-time and granular insights to make imperative business decisions. It enabled clients to-
- Use Automated Weekly forecast based on COVID-19 adjusted elasticities.
- Understand demand and consumer behavior during store closures. Updated “Corona elasticity” with different elasticities by channel ( .com vs store)
- Execute Inventory transfers based on new forecasts
- Fed new COVID adjusted forecast to the Allocation system to improve allocation.
The outcomes can be delineated as follows:
- +20M Savings through balance of year
- Overall forecasting error went from 40% to 10%
- Savings from trapped inventory transfers- $12M
- Savings from elimination of stockouts – $ 8M
IA’s AI-guided forecasting model “ADA” helped the client to decrease the forecasting error for a large specialty retailer from 40% to 10% in 3 weeks and added +20M savings to the bottom line.