Smarter Substitutions

 
 
 
 

Problem

Let’s say you ordered something online, but when the buyer goes to the store, it’s not available. How do you decide what the customer would substitute that item for?

There are a lot of factors that go into picking the correct substitute item and it’s different for each customer. If the wrong substitution was recommended or none at all, it creates a poor experience and leaves the customer unsatisfied.

Walmart uses an AI, called Smart Substitution, to offer alternative items when there’s a lack of supply and orders go unfulfilled. However, we need help from our merchants to increase performance, accuracy and strengthen the Smart Subs algorithm for future recommendations.

Currently merchants are unable to easily evaluate which items need their input and must reactively recommend items without having the right data available to make informed decisions.

Goal

Deliver an excellent online pickup and delivery experience, with assortment options that delight our customers, even when we don’t have the exact product that was ordered.

Because

Create a workflow that allows merchants to input substitution recommendations for items

to deliver a better experience by offering an assortment of options that delight our customers, even when we don’t have the exact product that was ordered.

Deliver an excellent online pickup and delivery experience by providing better recommended substitutions for items that are not available.

Allow merchants to easily identify and evaluate items that have high nill pick loss (when a customer paid for an item but had to be refunded because we could not fulfill the order). Also offer the ability to communicate back and forth with the Assortment and Smart Subs teams.

 

Background

Many merchants don’t know that a process exists. The current process is very manual where merchants interface via email and excel with a Merch Ops team. When there is a new substitution recommendation by the merchants, they are tested and if they pass a certain threshold of acceptance rate, they remain as substitutes and those that do not, drop off.

Currently there are many items that do not have recommended substitutes or have substitutions that have high rejection rates.

For this project the team consists of myself, product manager, business and engineering team. My primary responsibilities included creating initial proposal including requirements, redesign, and producing high fidelity mocks and prototype.

 
 

Design