Is this case study relevant to you?
Explore how our solutions can implement to your business
Optimization Models: How to Predict Conversion Rate and Net Revenue Effectively
Will a prospect respond more profitably to an offer of a premium incentive or a percentage discount?
Goal
6 models to predict the propensity to buy and net revenue
By sifting through your clickstream and transaction data, FocusKPI experts can uncover loyalty patterns and allow quick test/learn approaches to analyze different offers and scenarios.
Email Offers Data Sampling:
Data sources
-
3 year transaction history
-
3 year image upload history
-
2 year contact history
-
Census data
-
Campaign data not available for Jun- Sep 2012
Time windows: Use 24 months history to predict one week purchase
Two offer levels: High discount vs. Low discount
Three product categories:
-
Photo Book (High discount)
-
Photo Gift (High discount)
-
Cards (Low discount)
The model effectively both conversion rate and net revenue
-
Image upload recency and image upload quantity are strong indicators of conversion
-
Past purchase in PG, Prints, Cards … drives next purchase
-
Customer tenure since first-order indicates loyalty
-
Discount helps conversion
-
Past contact helps
Recommendation of Best Offer Scenarios
Hypothetical Example: 20 Customers/6 model scores
Maximize returns by giving customers the right products/offers
-
Email campaign: Photo Book 30% off. Total expected net revenue $110
-
Apply mathematical algorithms to maximize the net revenue constraining on offer quantities. Total expected net revenue of $166.
Slide right to view all tabs
Will a prospect respond more profitably to an offer of a premium incentive or a percentage discount?
Goal
6 models to predict the propensity to buy and net revenue
By sifting through your clickstream and transaction data, FocusKPI experts can uncover loyalty patterns and allow quick test/learn approaches to analyze different offers and scenarios.
Email Offers Data Sampling:
Data sources
-
3 year transaction history
-
3 year image upload history
-
2 year contact history
-
Census data
-
Campaign data not available for Jun- Sep 2012
Time windows: Use 24 months history to predict one week purchase
Two offer levels: High discount vs. Low discount
Three product categories:
-
Photo Book (High discount)
-
Photo Gift (High discount)
-
Cards (Low discount)
The model effectively both conversion rate and net revenue
-
Image upload recency and image upload quantity are strong indicators of conversion
-
Past purchase in PG, Prints, Cards … drives next purchase
-
Customer tenure since first-order indicates loyalty
-
Discount helps conversion
-
Past contact helps
Recommendation of Best Offer Scenarios
Hypothetical Example: 20 Customers/6 model scores
Maximize returns by giving customers the right products/offers
-
Email campaign: Photo Book 30% off. Total expected net revenue $110
-
Apply mathematical algorithms to maximize the net revenue constraining on offer quantities. Total expected net revenue of $166.