To help overcome demand planning challenges, such as statistical forecasting and promotion lift planning and reviews, we:
1. Implement data-driven collaborative planning
- Data-driven demand planning: Leverage data from various sources to help algorithms analyze vast historical data and detect subtle demand signals—such as weather patterns, social media posts etc. that may elude traditional analysis.
- Demand sensing: Actively monitor real-time market signals such as customer feedback, social media sentiment and point-of-sale data to adjust forecasts as needed.
- Collaborative planning: Involve multiple stakeholders in the process, such as suppliers, distributors, sales teams, and marketing teams to create a shared and accurate view of the demand.
2. Determine historical promotional lift
- Unified view: Collate data from different sources to create a combined view of past promotions with relevant information (for example, product, promotion timing, offer, promotion type, channel etc.).
One way to determine historical promotion lift can be to use parameters and statistical analysis:
- Parameters: Define promotion by a set of promotion parameters (for example, timing, discount, type etc.).
- Statistical analysis: Perform statistical analysis to determine seasonality and historical lift. Lift can be determined by comparing baseline demand for a product in a week against the actual sales that happened during that week.
3. Generate future promotional lift
- Use machine learning: For future promotions, we can generate a lift based on regression analysis or any advanced machine learning model of historical lifts for similar promotions. Suggested lift from the mechanism can then be used to review the lift coming from the marketing team on an exception basis.