The Challenge
A consumer goods company made pricing and promotion decisions based on intuition and historical precedent, not data-driven elasticity understanding. Heavy promotional spend had no systematic way to predict demand impact or profitability. Excel-based scenario analysis was limited to evaluating 5-10 scenarios manually, and the team didn't understand how demand responded to price changes or promotional mechanics.
Our Solution
End-to-end ML-based pricing and promotion optimisation system with three core capabilities:
1. Advanced ML Elasticity Modelling
Deployed gradient boosting and ensemble methods to model multiple elasticity types:
- Demand-price elasticity: How sales volume responds to price changes
- Cost-price elasticity: How cost structures constrain pricing decisions
- Cross-elasticities: Product interactions, competitive pricing, promotional mechanics
2. Automated Scenario Generation
Built framework generating thousands of pricing/promotion combinations across product portfolio, promotional mechanics, competitive responses, and channel-specific strategies.
3. Optimisation Algorithm Integration
Mathematical optimisation identifying best scenarios from generated set based on revenue, margin, or volume objectives with constraint handling.
Results
Scenario Expansion
Thousands vs. 5-10 manual scenarios
Elasticity Models
Demand, cost, and cross-elasticities
ROI Visibility
Clear promotion investment projections
Key Impact
- Commercial team shifted from gut-feel pricing to data-driven elasticity understanding
- Category managers gained confidence with clear ROI projections for promotions
- Optimisation revealed counter-intuitive strategies (smaller discounts yielding higher margin)
- Framework integrated into quarterly pricing and promotional planning process
- ML algorithms captured complex patterns not visible in traditional econometric models