Optimal Pricing at Grocery Stores
Revenue maximization via optimal pricing policies based on purchase history
Traditional revenue optimization relies on generic demand assumptions
Traditionally, pricing is done through a revenue optimization approach where models are built to predict demand as a function of price, and then an optimization model is solved to maximize the revenue and arrive at the optimal price.
This approach generally has practical limitations: either the demand estimation is too broad and not personalized to a particular grocery store, or there is not enough data from a store to estimate demand.
We use Optimal Policy Trees that leverage data from all grocery stores and use intrinsic features such as customer demographics to cluster stores and make optimal pricing recommendations.
This approach generally has practical limitations: either the demand estimation is too broad and not personalized to a particular grocery store, or there is not enough data from a store to estimate demand.
We use Optimal Policy Trees that leverage data from all grocery stores and use intrinsic features such as customer demographics to cluster stores and make optimal pricing recommendations.
Combine the power of complex demand modeling and interpretable policy learning
From historical transaction data, we model the purchase probability for any given price as function of demographics. The outcomes under each pricing option are then fed into our Optimal Policy Tree engine, which learns an interpretable tree that recommends the optimal price for each cohort.
In the example policy tree shown here, it uses age, income, household size, and homeownership to separate the households into cohorts where different prices lead to optimal revenue. As an example, it identifies that it is optimal to choose the highest price for households with two adults, no children and income above 150k.
In the example policy tree shown here, it uses age, income, household size, and homeownership to separate the households into cohorts where different prices lead to optimal revenue. As an example, it identifies that it is optimal to choose the highest price for households with two adults, no children and income above 150k.
Illustrative Optimal Policy Tree prescribing the prices that optimize revenue
Global optimality leads to increased revenue
We evaluated the performance of this policy tree and observed a 77% lift in revenue compared to the pricing observed in the data, which is a significant lift from both the current practice as well as other prescriptive tools in the literature.
Unique Advantage
Why is the Interpretable AI solution unique?
-
A novel approach to policy learning
Optimal Policy Trees are the first scalable and optimal solution to this class of problem
-
Flexible demand modeling
This framework allows for complex and accurate modeling of demand while maintaining interpretable policy recommendations