Promotion Analysis and Prediction from TPM data

3AI September 10, 2020

US subsidiary of one of the world’s largest food company

Problem Statement

• Company had promotion data for a specific retailer in TPM system and wished to leverage it for advanced analytics and forecasting
• Frequency of promotions were extremely high for the specific retailer, resulting in frequent dips of sales below the baseline
• The effect of frequent discounting required a new approach to sales response modeling and forecasting

Analytics Led Approach

• Conducted historical data review and provided insights, trends on promotion performance and sales/revenue distribution by chain and geography
• A unique methodology to model the effect of continuous discounting on future sales was developed. This resulted in very reliable models to explain promotion responses and volume dips below the baseline
• Incorporating this new approach provided insights into actual promotion tactic performances by retailer and geography

Business Impact

• Volume forecasts were generated using the customized model and benchmarked against actual values
• Analysis and model fitment provided insights into actual (decomposed) effectiveness of individual promotion tactics employed
• Generated sales forecasts to aid demand planning using customer’s promotion plan for the next year

Critical Success Factors

• Recommended a more optimal promotion strategy, based on variation in tactic (discounts, features and displays) effectiveness by retailer and geography