Whitepaper: Out of Stock Product Recommender

Solving the Peanut Butter Problem

The "peanut butter problem" occurs when a customer orders a specific kind a peanut butter to be delivered in a grocery order, but the picker fulfilling the order sees that it’s not on the shelf. Which of the fifty other kinds of peanut butter should the picker select instead?

This whitepaper lays out a framework for substituting out of stock items from the perspective of a retail store or order fulfillment platform.



We first define the problem from a business perspective and argue for how this solution helps a store’s bottom line. Then we propose a design for building a recommendation engine that uses available data to estimate a percent match between all products within a store. Finally, we propose an architecture for hosting the product data and recommender using a modern data warehouse.

Our proof of concept shows that the recommendation engine successfully identifies meaningful product substitutions. The design is robust to work with limited data and flexible to allow for customization at both the store level and customer level.

What you'll learn: 

  • How today's economic environment is putting a high strain on traditional stores and ecommerce
  • Technical challenges relating to supply chain disruptions
  • Identifying substitute products through predictive models
  • Business value of out-of-stock recommenders
  • Incorporating continuous customers feedback

Fill out the form to access your copy of Out of Stock Product Recommender whitepaper.


One of the most challenging aspects of translating an in-store experience to an online experience
centers around product and inventory out of stocks.

- Steve Krause, BlueGranite Retail Lead