The algorithm that identifies cross-sell products is based on data about products that were purchased together and products with similar descriptions. It uses this data to provide a mix of those related products as recommendations for a given product.
In cases where purchase or product description data aren't available for a given product, products from a related collection are shown. The algorithm uses the collection in the URL from which the request to the API endpoint originated, and pulls other products from that collection. If the product doesn't have a collection URL, then the recommendation algorithm finds collections that include the product and pulls other products from those instead (excluding collections with handles “all” and “frontpage”).
The recommendation algorithm associates up to ten products per individual product, in order of relevance. For example, the first cross-sell product is more relevant than the tenth cross-sells product. Because of this, it's a good idea to show no more than four products per product page to promote only the most relevant recommendations.
Requirements and limitations
The types of product recommendations that appear are determined by your online store and your Shopify plan:
Purchase history and product descriptions: Plus plan and having <7000 products published in your Online Store, with an English storefront
Purchase history only: Plus plan and having <7000 products published in your Online Store, with a non-English storefront
Purchase history only: Non-Plus plan and having <7000 products published in your Online Store
Collections: Any plan with >7000 products published in your Online Store
The recommendation algorithm doesn't use orders that have been imported from another store or e-commerce platform to generate product recommendations.
Products that are out of stock, set to a price of 0, or gift cards aren't included in recommendations.
You can use the Boostheme Toolkit app to specify the products as recommendations for a given product. Please view more details here.