The Business Problem
In the CPG industry, inaccurate demand forecasting leads to two costly problems: stockouts (lost sales) and overstocks (wasted capital and expired product). Traditional forecasting often relies on simple historical averages and fails to capture complex patterns like holidays, seasonality, and recent trends.
The AI Solution
This project showcases a predictive analytics tool that uses a time-series AI model (Meta’s Prophet) to generate accurate, SKU-level demand forecasts.
The model was trained on a hyper-realistic, multi-year dataset that includes patterns for:
- Weekly and yearly seasonality
- Holiday spikes
- The impact of promotions
- The momentum of “New Era” vs. “Traditional” products
Below is a short video walkthrough demonstrating the application’s capabilities.
Coming soon
The tool provides not only a visual forecast but also a set of actionable business metrics, including the expected sales volume for the next 30, 60, and 90 days, and an analysis of the product’s underlying growth trend.