Personalized Product Recommendation

This guide explains all configuration options available in the Personalized Product Recommendation module. By adjusting these settings, you can create tailored product recommendations that align with your business goals while providing your customers with relevant recommendations. The guide walks you through the implementation workflow and details each configurable property with its purpose, available options, and common use cases.

Implementation Workflow

Design Phase: Develop a variant or campaign in preview mode and set the necessary business rules in the 'Controls' section. Some setup settings are necessary:
- Cron: This is an expert property that must be set by a CM professional. A cron must be defined to determine the frequency and timing of the personalized product recommendations. This can be performed using the Cronbuilder tool. The Cronbuilder facilitates cron job creation, allowing users to specify exact times and days for the recommendation process to run.
- Delivery Method: This is an expert property that must be set by a CM professional. It can be either:
  - ABS: When selected, a custom data delivery configuration is set up by a CM professional.
  - CDP: When selected, requires the following additional configuration parameters:
    - Tenant ID: Available in the CDP platform under Settings > API tokens.
    - Event Type ID: Navigate to CDP > Internal Sources > Raw Sources. Search for the right ID
    - Access Token: Accessible in the CDP platform under Settings > API tokens.
Refinement Process: Evaluate performance by experimenting with different business rule configurations in the cockpit
Publish: When satisfied with the configuration, publish the variant or campaign to implement the solution in the live environment

Configurable Properties

Below are the configuration properties organized by category to help you create the right recommendations for your customers. These properties can be found in the Cockpit.

Basic Configurations

These fundamental settings control the core recommendation functionality, including how many products are shown and what happens when personalization data is limited.

Product Selection Rules

These filters determine which products can appear in recommendations, allowing you to exclude certain items (like out-of-stock products) and focus on promoting others that align with your business strategy.

Ranking Modifications

These options let you adjust recommendation order by prioritizing specific items based on their attributes, including the ability to place specific products in guaranteed positions.

Diversity Control

These settings ensure customers receive varied and relevant recommendations by preventing repetition, balancing different product types and managing the mix of previously purchased items.

Campaign Properties

Note: Campaigns are a different solution type with only the two properties listed below. These are separate from the main recommendation variants and have a simplified configuration.

1. General Settings

Number of Recommendations

Purpose: Determines how many recommendations each customer will receive
Options: 1-50 (integer)

Rotation of Recommendations

This is an expert property, which means that it needs to be set by a CM professional.
This setting determines how often recommendations rotate through your list. When Rotation of Recommendation is set to 3, the system selects items at positions 1, 4, 7, etc. (starting at position 1 and skipping 2 positions each time). This approach ensures users see high-ranking items during subsequent recommendation cycles rather than working strictly sequentially through the list.
Required value: 1-5 (integer)

Insufficient Recommendations Threshold

This is an expert property, which means that it needs to be set by a CM professional.

Maximum allowed percentage of users with too few personalized recommendations
Required value: 0-100 (percentage)

2. Email Filter

Purpose: Determines if recommendations should be restricted to users with email addresses
Options: Enable/Disable (boolean)

3. Fallback Configuration

Fallback Method

Purpose: Provides recommendations when personalized options are insufficient
Options: Currently only "popularity" is supported
When Used: When a user has insufficient personalized recommendations it is filled with popular recommended items.

Number of Fallback Recommendations

Purpose: Sets how many fallback items to provide per user
Options: 1-1000 (integer)

4. Item Exclusion Criteria

Purpose: Prevents specific items from appearing in recommendations
Options:
- Enable/Disable: Turn this filter on or off
- Exclusion Criteria: SQL expression builder defining which items to exclude (e.g., category = 'Bread')
Common Uses: Excluding out-of-stock items or low-margin products

5. Item Filter

Purpose: Restricts which items can appear in recommendations
Options:
- Enable/Disable: Turn this filter on or off
- Filter Criteria: SQL expression builder defining which items to include (e.g., collection = 'Women')
Common Uses: Ensuring only Women's collection items appear when sending targeted email campaigns to female customers.

6. Boost By Numerical Feature

Purpose: Increases ranking of items based on numerical attributes
Options:
- Enable/Disable: Turn this boosting on or off
- Boost Feature: Select a numerical attribute (e.g., profit_margin, price)
- Boost Strength: Value between 0-10 that determines boost intensity
How It Works: Higher values of the selected feature receive stronger boosts in rankings
Use Case: Promoting high-margin products or adjusting for price points

7. Boost By Condition

Purpose: Increases ranking of items matching specific category values
Options:
- Enable/Disable: Turn this boosting on or off
- Boost Condition: SQL expression builder defining which items to boost
- Boost Strength: Value between 0-10 that determines boost intensity
Example: Boost items where is_new_arrival = true with strength 5.0
Use Case: Promoting new arrivals, featured items, or specific brands

8. Fixed Position Recommendations

Purpose: Places specific items at guaranteed positions in recommendation lists
Options:
- Enable/Disable: Turn this positioning on or off
- Item Condition: SQL expression builder identifying which item to position
- Target Position: Fixed position (1-15) for the chosen item
How It Works: Rules are applied in order, with each rule affecting only one item
Use Case: Ensuring promotional items appear in specific positions of recommendations

9. Exclude Purchased Items

Purpose: Prevents recommending items the user has already purchased
Methods:
- all: Exclude all previously purchased items
- recent: Exclude items purchased within a specified time period
- recentExceptEngaged: Exclude recent purchases unless the user has engaged with them again
- recentAllowXBoughtItems: Allow a specified number of previously purchased items
Settings:
- Similarity Feature: Product feature to determine which similar items to exclude
- Exclusion Period: Number of days (1-365) to exclude purchased items
- Item Filter: SQL expression builder for including/excluding items from this rule
- Allowed Purchased Items: Number of highest-ranked purchased items to include (0-15)
- Engagement Exceptions: Engagement types that override exclusion (add-to-cart, view, etc.)
Use Case: Avoiding redundant recommendations while allowing repurchase of consumables

10. Exclude Previously Recommended Items

Purpose: Prevents repeating recent recommendations to provide variety
Methods:
- all: Exclude all previously recommended items
- recent: Exclude items recommended within a specified time period
- recentExceptEngaged: Exclude recent recommendations unless the user has engaged with them
Settings:
- Similarity Feature: Product feature to determine which similar items to exclude
- Exclusion Period: Number of days (1-365) to exclude recommended items
- Item Filter: SQL expression builder for including/excluding items from this rule
- Top Recommendations to Exclude: Number of top previously recommended items to exclude (1-15)
- Engagement Exceptions: Engagement types that override exclusion
- Personalization Selection: Which personalization outputs to exclude from current recommendations
Use Case: Ensuring customers are not recommended the same items over and over again.

11. Limit Similar Recommendations

Purpose: Restricts number of recommendations from the same category for better variety
Options:
- Enable/Disable: Turn this diversity control on or off
- Grouping Feature: Product feature to group similar items (e.g., Category, Brand, Color)
- Maximum Similar Products: Maximum number of products allowed with same feature value (1-15)
Example: Limit to no more than 3 products from the same category
Use Case: Ensuring customers see a diverse range of products

12. Limit Products By Specific Features

Purpose: Controls product mix based on specific attributes
Options:
- Enable/Disable: Turn this feature limit on or off
- Product Feature Condition: SQL expression builder defining product feature values to limit
- Maximum Items: Maximum number of items to include with this feature (1-15)
Example: Limit products where category = 'shorts' to maximum 2 items
Use Case: Controlling recommendation mix to align with business goals

13. Campaign Item Filter

Purpose: Temporarily filters items specifically for marketing campaigns
Options:
- Enable/Disable: Turn campaign filtering on or off
- Campaign Filter Expression: SQL expression builder defining which items to include in campaign
Use Case: Promoting seasonal collections, new arrivals, or special events

14. Campaign Item Boost

Purpose: Temporarily increases ranking of specific items during campaigns
Options:
- Enable/Disable: Turn campaign boosting on or off
- Boost Condition: SQL expression builder defining which items to boost
- Boost Strength: Value between 0-10 that determines boost intensity
Use Case: Ensuring promotional or seasonal items receive higher visibility