Configure a Recommender for Product Recommendations

When you create a product recommender, you specify the type of recommendation and the strategy you want to use. You can also add rules to fine-tune the recommendation.

Before you can configure a recommender for product recommendations, make sure that an administrator has configured catalog and order feeds for your site. These feeds provide Einstein Configurator with necessary information for configuring the recommenders. Also make sure that the administrator sets up an account for you to use the Einstein Configurator tool.

To configure a recommender:

  1. In a web browser, access Einstein Configurator.
  2. Log in to Einstein Configurator using the credentials provided to you by your administrator.
  3. From the Select Site... list, select your site.
  4. Click the Site Recommendations tab.
  5. Click New Recommender
    The Create New Recommender dialog opens.
  6. In the Choose a Recommender Type section, select one of the recommender types.
  7. In the Recommender Name field, enter the name of the recommender.
    Note: The recommender name must be unique for the site. The name can contain letters, the hyphen (-) character, and the underscore (_) character. The length of the name must not exceed 100 characters.

    You can't change the recommender name after you save it. The recommender name appears in other locations, for example, the Content Slot recommender and in Einstein Dashboard reporting. Best practice is to include the name of the page where you plan to apply the recommender as part of the recommender name.

  8. Click OK.
    The Create New Recommender dialog closes, and the Recommender Configuration page opens.
    Note: The newly created recommender isn't saved at this point. It's saved later in this procedure when you click Save.
  9. (Optional) In the General tab:
    1. Select an upper limit on the number of recommendations in the Max # of Recommendations field.
    2. Enter a description for the recommender.
      This description is visible within the Business Manager > Content Slot > Recommender selection.
    3. Enable (check) the Don't recommend products purchased option and select a value from the dropdown menu.

      This option enables you to define how many days (7, 30, 60, or 90) a recommender waits before recommending previously purchased products. The default setting is 30 days.

    4. Select up to two products that you want to pin to the recommender.
      Pinned products always appear as top results for the recommender.
    5. Specify an entity type of either base product or Variation Group.
      You use the Entity Type setting when the product catalog contains variation groups. The setting enables you to service recommendations at the base product or variation group level.
  10. (Optional) Click the Strategies tab.
    Note: This tab isn't available for all recommenders.
    The Strategies tab shows a table where each row is a single strategy. The highest priority strategy appears in the top row, the second highest strategy, if any, appears in the second row, and so on. You can reorder strategies using the up and down arrow icons in each strategy row.
  11. (Optional) To create another strategy, click Add Strategy.
    A strategy row appears at the bottom of the table. You can then select a new strategy by choosing a value in the Strategy Name field and clicking Save.

    To delete a strategy, click its row in the table and then click the trashcan (delete) icon.

  12. (Optional) Click the Rules tab.
    Recommender rules enable control over what products are shown to shoppers.
  13. (Optional) Click Add Rule.
    A new row appears in the rules table where you can create a rule by selecting values and clicking Save.
  14. To save the configuration, click Save in the upper right-hand corner of the page.
    Note: If you click Cancel, or if you leave the configuration page entirely, your changes are discarded.
While you’re configuring strategies and rules, you can preview them to see what recommendations are passed to the storefront at run time.