Skip to main content

Recommendations

Recommendations are suggestions for complementary items or replacement items. Customers check out their carts and shoppers facilitate item replacements every day. This customer feedback provides a rich understanding of what makes a great replacement or complementary option for an item. The Instacart recommendations service uses machine learning to analyze purchasing patterns across all orders fulfilled by Instacart and to rank items based on popularity and other factors. This data is used to improve the relevancy of recommended items.

Complementary items​

Complementary items are the types of items that customers often buy together in a cart. For example, if a cart contains peanut butter, your storefront might display a list of complementary items, such as a selection of jams, jellies, marmalades, bananas, and bread. A well-curated list of complementary items can improve the customer’s shopping experience and might increase the number of items in the customer’s cart.

The Instacart recommendations service uses machine learning to rank items based on popularity, and then generates a list of popular items that are complementary to the items in the cart. Based on where in the shopping journey the list is to be displayed, other factors might be used to adjust the list.

When displaying complementary items, the following rules apply:

  • Items must be available in the selected store.
  • If a complementary item is currently in the customer's cart, it is omitted from the list.
  • Items are displayed in order of popularity, unless other tools are used to boost items to the beginning of the list.

A customer's previous purchases can be a factor when generating complementary items, but it depends on where the complementary items show up in the user journey. For more information, see the Related topics.

Replacement items​

Replacement items are the items that customers would see on the shelves surrounding a preferred item while shopping in store. You can provide a similar experience for online customers by displaying recommended replacement items. High quality recommendations for replacement items increase the chance that your customers can buy the types of items they need, even if the items aren’t their first choice of brand or size. From both the customer and retailer perspectives, a satisfactory replacement is always preferable to a refund.

For each item in a cart, the Instacart recommendation service uses machine learning to generate a list of replacement items.

The factors used to select the replacement items for a cart item can include things like:

  • An item's similarity to the cart item.
  • How often other customers accepted an item as a replacement for the cart item.
  • The customer’s replacement history for the cart item.

When displaying replacement items, the following rules apply:

  • Items must be available in the selected store.
  • A customer's past replacement selections are shown at the beginning of the list.
  • All other items are shown in order of popularity.

Ads​

Sponsored product ads can appear in some of the complementary items lists. Some of the sponsored products are boosted to the beginning of the list. Other sponsored products may be interspersed among the items in the list. The number and placement of the ads are subject to change without notice. For more information about the ads service and where ads appear in the storefront, see Ads.