Create A Premium Recommendation

A recommendation is a combination of criteria and priority settings that are used to created a targeted list of products from your product catalog. A premium recommendation uses predictive models or filters based on a reference contact or reference products in order to create highly-personalized product recommendations. You can build a standard recommendation using either Recommendations Standard or Premium, but you must have Recommendations Premium to create a premium recommendation.

About this task

Building a premium recommendation is broken into six functional areas:

  • General Settings are the configuration options for the recommendation.
  • Product Criteria are the rules that you apply to your product data in order to determine which products qualify for the recommendation.
  • Related Products Criteria are filters that can limit the recommendation results based on how products in your catalog relate to a reference product.
  • Contact History Criteria are filters that can limit the recommendation results based on contacts' order and browse data.
  • Priorities determine how the products that meet the product criteria settings are sorted and ranked against each other.
  • Reference Products are settings that can be used to identify how Bronto should determine the reference products used for a recommendation.

When you create your recommendation, the changes you make are automatically saved as a draft version of the recommendation. This allows you to use the Preview Results button to view the results of your recommendation as you build it so you can adjust your recommendation until you're happy with your results. If you exit the recommendation while you are creating it without publishing it, all of the settings, priorities, and criteria you created and saved will be kept as a draft.

Once created, this can be used to design dynamic emails that will always include the latest, most relevant products from the recommendation you use in your message.

To create a recommendation:

Procedure

  1. In Recommendations, navigate to the Recommendations page by clicking the icon.
  2. Click New Recommendation.

    The Create Recommendation window is displayed.

  3. Enter a name for your recommendation in the Label box.

    The name can only contain alphanumeric characters.

  4. If desired, turn Predictive Model to On and select a predictive model to use for the recommendation.

    Bronto's predictive models use patterns of behavior related to a reference product in order to provide recommendation results. You don't have to use a predictive model to create a personalized recommendation. The available predictive models are:

    • Bought This Bought That returns a set of products typically bought by customers who bought the reference product.
    • Frequently Bought Together returns a set of products typically bought by customers in the same order in which they bought the reference product.
    • Browsed This Browsed That returns a set of products typically browsed by customers who browsed the reference product.
    • Browsed This Bought That returns a set of products typically bought by customers who browsed the reference product.
    • Best Product returns a set of products the reference contact would be most interested in using the contact’s 10 most recently browsed products and 10 most recently ordered products as reference products.
    • Similar Content Products returns a set of products that have similar words in the title, description, brand and product category fields (if mapped). For this recommendation to be effective, if your site has multiple variations of the same base product in your product catalog, you will want to leverage the Parent Product ID field to group these products and enable the Restrict Parent/Variation Results to option on the recommendation so your results aren’t monopolized by these products based on their near-identical product content.

    You cannot change a recommendation's predictive model setting once you create it. The results of predictive models are recalculated nightly.

  5. If desired, enter a brief description for the recommendation in the Description box.
  6. Click Create Recommendation.

    The Recommendation Settings page is displayed.

  7. If you are not already on the General Settings tab, click General Settings .
    1. Select a recommendation from the Use Backup list if you want to use another recommendation you created if this one doesn't return enough relevant results.

      This is a list of the active standard recommendations you have built. If you do not have any other recommendations or do not want to use a backup, select Don't Use.

    2. Set Use Default to On if you want to use the default backup recommendation you configured for your account.

      If this is set to on, the default recommendations are only used when your recommendation (and backup recommendations) do not provide the number of desired results.

  8. Click Product Criteria to open the Product Criteria tab.
  9. Select an option for Restrict Parent/Variation Results to

    In order to use this, you must use the Parent Product ID field in your product catalog When a product has a parent and multiple variations of that parent, this setting will determine which product results to use in your recommendation. For example, you might want to limit a result to One Variation Only if the variant products are very similar and might result in what appears to be duplicate recommendations. (Items of a same color, for example.)

    No matter what you select any product that isn't a parent or variation will be included in the recommendation results if it matches the other criteria you set for the recommendation.

    • Parent Only if you want your recommendation to only include parent products that match the recommendation's criteria.
    • One Variation Only if you want to only include the variation product that best matches the recommendation's criteria.
    • Multiple Variations Only if you want to include the variation products that match the recommendation's criteria.
    • Parent and Multiple Variations if you want to include any product, parent or variation, that matches the recommendation's criteria.
  10. Set Set Inventory Minimums to On if you want to ensure that only products with a certain amount of stock are returned.

    This setting uses the value stored in the Inventory Threshold and Quantity product fields. Only products that have a Quantity amount that is equal to or greater than the value in the Inventory Threshold will be returned when Set Inventory Minimums is set to On.

  11. Set Use Global Criteria to On if you want to use the default criteria you configured for your account.

    When you set this to on, the Global Criteria you have set up are displayed on this page.

  12. If you want to include a single criteria setting that you can customize within the Rec Loop of a message:

    Use Request Time Criteria lets you reuse a recommendation for multiple purposes by providing a single, dynamic field that can be modified at the message level. You will set the field and operator here, and then provide the value you want to use within the message. For example, you could build a recommendation where Use Request Time Criteria is set to Product Category equals and then use the recommendation to feature different product categories in different emails.

    1. Set Use Request Time Criteria to On.
    2. Select a field from the first list.

      See either product, commerce, or browse field help topics for more information about what each of these fields are.

    3. Select an operator from the second list.

      The Operators you see are determined by the type of Field you selected. See step 15 c for more information about each type of operator.

    4. When you use this recommendation in a message, map the value you want to use within the Rec Loop.

      See Recommendation Messages for information on how to do this. If you set Use Request Time Criteria in your recommendation but do not set a value in an email that uses the recommendation then this field will be ignored.

  13. Click add, located to the right of Criteria, to set up the criteria used in the recommendation:
    1. Select a field to use from the Field list.

      See either product, commerce, or browse fields for more information about what each of these fields are.

    2. Select the rule you want to apply to the field from the Operator list.
    3. Enter or select the value you want the criteria to meet in the Value box.

      The Operators and Values you see are determined by the type of Field you selected. While there are several types of fields, these fields fall into three main categories:

      • Date-based. When you select a date-based field, you can specify either an exact date or a range of days relative to the operator you selected. For example, you can set a date range within the next week by selecting On or Before and entering 7 for the days from now value.
      • Numeric. When you select a numeric field, you can specify an exact number or a number range that is relative to your Operator.
      • Text-based. There are a few rules for text-based fields, which are basically any field that isn't a date or number. If your Operator is set to Equals or Does Not Equal, the content you enter for Value must be an exact match for the information stored in the field, including capitalization, and should not be wrapped in quotes. For the operators In, Not In, Contains, or Does Not Contain the values entered must be provided in a comma-separated list and each individual value must be in quotes. For example: "Sneakers", "Shoes, High tops", "Shoes, Hightops", "Shoes, High-tops", "Tennis shoes". In this example, the comma between shoes and high tops is included as part of the field in the database. The use of quotes lets Recommendations identify the difference between this type of comma and a comma separating items in your list of Values. The quotes are always required, even if there is only one item in your list.

      Some Field and Operator combinations may not require a Value.

      After you enter your value, you can click Recommendations Get Product Count to see how many products in your catalog match the criteria you specified.

    4. Click Save & Close to save your criteria and move on or Save & Add Another to save your criteria and add another global criteria option.
      Note: You still have to click Publish Changes for these criteria to be available for your recommendations.

    For example, if you are creating a recommendation for items targeted to women, Field is set to Gender, Operator is set to Equals, and the Value is the word your product data uses to indicate it is a woman's item.

  14. Click the Related Products Criteria tab to filter the recommendation results based on how products in your catalog relate to a reference product:
    1. Set Same Brand/Category to On in order to include or exclude products from the same brand or category as the reference product.
      These options are only available if you have the brand and category product fields mapped in your product catalog. Select:
      • Exclude Products from the Same Brand to exclude products that have the same brand in your product catalog from being included as a recommendation.
      • Limit Products to the Same Brand to limit recommendations to products that have the same brand in your product catalog.
      • Exclude Products from the Same Category to exclude products that have the same product category in your product catalog from being included as a recommendation.
      • Limit Products to the Same Category to limit recommendations to products that have the same product category in your product catalog.
    2. Set Specified Products to On in order to either limit the results to items that are whitelisted in your product catalog or exclude items that are blacklisted in your product catalog.
      For the given reference products, the entire list of whitelist or blacklist products will be consolidated, and the first 250 random products will be used. The other products will be ignored from the criteria.
      Note: These options are only available if you have the product whitelist and product blacklist fields mapped in your product catalog.
    3. Set Similarly Priced to On in order to set a price range recommended products should fall into.
      This price range will be within a percentage of the reference product. Select:
      • Within to have recommendations be items that are priced no lower or higher than the set percentage.
      • Greater Than to have recommendations be items that are priced equal to or up to the indicated percentage higher than the reference product.
      • Less Than to have recommendations be items that are priced equal to or up to the indicated percentage lower than the reference product.
      then enter the percentage within the box. For example, you select Less Than and set the percentage to 10%. If the reference product costs $100 then all recommendations will be between $90 - $100.
  15. Click the Contact History tab to set filter criteria based on a contact's order and browse behavior.

    These settings will use the order and browse data for the contact who is sent the message.

    1. To prevent a contact from seeing a recommendation for an item they recently ordered set Exclude Ordered Products to On and indicate how many days back you want order data to be considered.
      Up to the last 250 products ordered will be excluded within the time period specified by the user.
    2. To make sure only products a contact recently browsed are in the recommendation set Limit Results to Browsed Products to On and indicate how many days back you want browse data to be considered.
      Up to the last 250 products browsed will be excluded within the time period specified by the user.
  16. Click the Priorities tab to define your product priorities:
    Note: Priorities tab is not available when you have a predictive model associated with your recommendation.
    When setting priorities, do not use the same values you used for your criteria settings. Instead, identify how you want to rank the products that match your recommendation criteria against each other. Each recommendation can have no more than three priorities. If you try to add more than three, you will be asked to delete one of the existing priorities before adding a new one.
    1. Choose an option from the Apply Randomness to Results dropdown.
      • Do Not Use- This is the default option.
      • Randomize Best Results- Products that meet both your recommendation's criteria and priorities will be randomized.
      • Randomize All Results- All of the products that meet the recommendation's criteria will be randomized.
        Note: If you select this option, you can not add a priority.
    2. Click add located to the right of Priorities.
    3. Select a field to use from the Field list.

      For Recommendations Standard or Premium you can select product fields and for Recommendations Premium you can also select commerce or browse fields.

    4. Select the rule you want to apply to the field from the Operator list.
    5. Enter or select the value you want the criteria to meet in the Value box.

      All of the rules outlined in the step regarding criteria also apply to the Field, Operator, and Value options for priority. Additionally, you have the option to select ranking-based values from the Operator list. For example, you can set a date from newest to oldest or a numerical value from lowest to highest, etc.

    6. Select the priority from the Priority Level list.
      Priority is a way of creating a hierarchy of products within your recommendation results:
      • Strongly Upgrade Products that match the priority settings are given a significant boost in the results. For example, if a priority is based on Price Highest to Lowest, the highest priced items are ranked highest in the results and the lowest priced items are not boosted in the rankings.
      • Slightly Upgrade Products that match the priority settings are given a boost in the recommendation results, but the boost is less than those with Strongly Upgrade.
      • Slightly Downgrade Products that match the priority settings are given a downgrade in their relevance in the results.
      • Strongly Downgrade Products that match the priority settings are given a significant downgrade in their relevance in the results. For example, if a priority is based on Price Highest to Lowest, the highest priced items are ranked much lower in the results, and the lowest priced items are not given a downgrade.

    So, if your criteria is women's items and you want to make sure you recommend items you have a large stock of, you can set up the priority with Field set to Quantity, Operator set to Highest to Lowest, and Priority Level set to Strong Upgrade in order to make sure items with a large quantity are surfaced higher in the list of recommendations.

  17. Click the Reference Products tab to filter recommendation results based on recent contact behavior.

    The options you set on this tab determine what the reference products are, not which products are included in the recommendation results. If you set reference products here then any products in your email will not be used as reference products.

    1. Turn Recently Browsed Products to On in order to base recommendation results on the last 10 products a contact has looked at.
    2. Turn Recently Ordered Products to On and select a date range in order to base recommendation results on the last 10 products a contact bought within that many days.
  18. To preview the results of your recommendation, click the Preview Results button.

    This provides a preview of the products that match the recommendation criteria and settings. If you have settings that are based on a reference product or reference contact, you will need to enter an example contact email address and/or Product ID in order to see the results of your recommendation.

    1. Enter a contact's email address in the Contact Reference box.
    2. Enter a product ID in the Product Reference box.

      If you have set criteria on the Reference Products tab you will not need to set a product ID.

    3. Click Preview.

    The Preview list can be switched between a list view Recommendations List Preview Button or an card view Recommendations Expanded Preview Button. Each view provides a slightly different overview of the products. Click on any product name to see the complete details for the product. Here you can see the number and percentage of products that matched your recommendation.

    If you update your recommendation with new criteria or priority settings then return to the Preview tab you will see these updates reflected in the recommendations on this page.

  19. Optional: If you are finished building the recommendation, click Publish Draft to make the recommendation available for use.

    You do not have to click Publish Draft to save your recommendation.