The dataminers from Delhaize demonstrated that personalized coupon recommendations optimised with TIMi are generating 4 times more visits in the store than unpersonalized folders.
Multi products assignment
Let’s assume that we need to send two promotions to each customer amongst a choice of 60 promotions currently available. The promotions (“coupons”) are typically a reduction on the purchase price of an item. The usual algorithms to assign the coupons won’t work here because we have a limited quantity of each coupon. We need to take this into account in order to compute the assignment that maximizes the ROI.
Classical solutions have an assignment solver that can compute the optimal assignment in the case where we need to assign one coupon to each customer (i.e. mono product assignment problem). Anatella can compute the optimal assignment in the case where we need to assign several coupons to each customer. The assignment problem often appears when you are deploying cross-selling predictive models (next-to-buy, next-best-activity, next-best-offer, next-best-product). The multi-product assignment is thus a critical functionality of Anatella.