Explore Einstein Product Recommendations

Brandon Wilson is a merchandiser at Cloud Kicks, a high-end sneaker company. He wants to improve the shopper experience on the Cloud Kicks ecommerce storefront. One way he can do this is to provide personalized product recommendations based on a shopper’s interests and actions. Another way is to track general shopping trends so he can recommend current popular products.

It’s a lot to consider for thousands of customers! Luckily, Salesforce has just the tool for him, Commerce Cloud Einstein.

The Cloud Kicks storefront runs on Salesforce B2C Commerce. Currently, Brandon uses Business Manager, the online tool for managing B2C Commerce storefronts, to configure explicit recommendations. Explicit recommendations are links from a source object, such as a category, product, or a product set, to a target object, such as a product or product set. When a shopper looks at a pair of red high-top sneakers on the Cloud Kicks storefront, for example, Brandon has configured a T-shirt and red shorts recommendation to show in the You Might Also Like area on the page.

Cross-sells (related products) or up-sells (more expensive products) are some typical explicit recommendations. Brandon already configured multiple types of cross-sells, which he categorized as a family of products, complementary or accessory products, and competing products. He shows them in the storefront under labels such as What's New, May We Also Recommend, You Might Also Like, and Best Sellers.

Explicit recommendations are based on a best guess (and sometimes wishful thinking). They are what Brandon and his team consider obvious connections. But without direct shopper input or feedback or live retail data, Brandon wonders if Cloud Kicks is missing sales opportunities. He wants to take the guesswork out of product recommendations by using leading-edge data science. Personalized product recommendations can also help Cloud Kicks maximize conversion rates and increase average order value.

That’s where Commerce Cloud Einstein Product Recommendations comes in!

Commerce Cloud Einstein dynamically generates recommendations using live data. You can think of them as implicit recommendations, in contrast to the explicit recommendations that Brandon configures in Business Manager. Not only does Commerce Cloud Einstein use historical and live data from shopper activity on your storefront, but it also uses data from all the merchants who agree to share their data—a much bigger pool. The more data that Einstein uses, the better informed its recommendations are for your shoppers.

To implement Einstein Product Recommendations, Brandon realizes early on that he needs help from both Linda Rosenberg, the administrator at Cloud Kicks, and Vijay Lahiri, the developer.

Here are the steps this team must take.

1. Linda configures and runs a catalog feed against the production instance.
2. Linda configures and runs an order feed against the production instance.
3. Linda deploys the Configurator tool.
4. Brandon configures recommenders (business rules) with Configurator.
5. Vijay creates a content slot in an Internet Store Markup Language (ISML) template.
6. Brandon creates a content slot configuration in Business Manager that controls the template and the business rules that Einstein applies to the recommended product IDs.

If you earned the Commerce Cloud Einstein Implementation badge, you followed along as Linda completed the first two tasks. If you earned the Smarter Search with Commerce Cloud Einstein badge, you learned how the team completed the third task. In this module, you see how they take the remaining steps to set up Einstein Product Recommendations.

Let’s get started on the fourth step, configuring recommenders. Recommenders provide the link between what the shopper views on the storefront and a list of recommended products.

Brandon is eager to learn how to create recommenders using the Configurator tool. This tool lets him control the strategy (algorithm) that Einstein uses to generate the list of product IDs, and it lets him create rules that control which products are shown, hidden, promoted, and demoted. When the storefront renders the final list of recommended products, they’re in the optimal order as defined by his rules.

Later, he works with Vijay on the content slot that he uses to render the personalized recommendations on the storefront.

First things first, Brandon steps back to plan his approach for using Einstein Product Recommendations. He answers these questions.

• Where will the recommendations live on a page?
• What types of pages will offer recommendations?
• What type of products will display?

Where he places recommendations is important. He wants them in a location that grabs shoppers’ attention, for example, beneath product details or in a left frame on the home page. The placement should always enhance what the shopper is already viewing. Remember, you’re linking what the shopper is currently viewing to products that might also interest them.

Brandon identifies which types of pages are best for recommending products. He can configure recommenders for just one page type or for all pages. Here are the types of pages he considers.

• Product details page
• Category pages
• Home page
• My Account page
• My Recommendations page
• Cart
• Wishlist
• Checkout

Recommendations should pique shopper interest or offer a broader solution or selection. Here are some examples.

Now that Brandon has an idea of the page types, page placements, and products that he wants to focus on, he researches what recommenders are all about. He learns that configuring recommenders involves assigning types, strategies, and rules. He must assign a type to each recommender, and he can assign up to three strategies and up to 30 rules per recommender.

Here are the recommender types.

• Product to Product
• Products in All Categories
• Products in a Category
• Recently Viewed
• Complete the Set

Here are the available strategies.

• Customer recently viewed items
• Customers who bought also bought
• Customers who viewed also viewed
• Customers who viewed ultimately bought
• Product Affinity Algorithm
• Real-Time Personalized Recommendations
• Recent Most-Viewed Products
• Recent Top-Selling Products

This illustration shows a Product to Product recommender that’s configured with two strategies and two rules.

Brandon Wilson learned how Einstein Product Recommendations can help improve the shopper experience with personalized product recommendations. He explored how Product Recommendations work and how to plan for and implement them.

In the next unit, Brandon learns more about recommenders and uses the Configurator tool to create them.

Salesforce Help: Product Recommendations