This article describes a new feature which is in early access and not available for general use at this time. This will be available in a future release.
If you are interested in being an early access client, reach out to your Customer Success Manager.
Whether you use a pre-made template or create a new one for your email campaign, you can add and customize your product blocks.
The product recommendation rules let you show different products to different customers based on their behaviors and preferences. This personalization ensures each customer receives the most relevant product recommendations tailored to their individual shopping habits and interests.
These rules are powered by an algorithm-based recommendation engine that analyzes customer behaviors including browsing history, purchases, and product interactions. Each recommendation type uses different data models to display the most relevant products to the right customers.
Recommendation rules
| Recommendation rule | Description | Configuration notes | Technical guides |
|---|---|---|---|
| Next Best Purchase | Displays products predicting a customer’s next purchase | No configuration available. | Next Best Purchase Recommendations Technical Overview |
| Recent Interaction History | Products the customer interacted with in the last 30 days. | Configurable to be based on views, add to carts, purchases, searches, or browsing history | |
| Co-Recommendations | Products explicitly based on customer’s actions | Configurable to be based on views, add to carts, purchases, searches, or browsing history | Co-Recommendations Technical Overview |
| Best Sellers | Products with the highest purchase count within the lookback window | Configurable lookback windows of 24 hours, 7 days, or 30 days | |
| New Arrivals | Products that were added to your Product Catalog within the lookback window | Configurable lookback windows of 24 hours, 7 days, or 30 days | |
| Dynamic Products from Catalog | Displays products based on applied product filters | Set one or multiple criteria for the block, such as On sale is true. |
Fallback
When an email is being sent to customers and the described criteria are not met, you can choose a fallback option.
You can choose multiple fallback options, including different recommendation rules, hiding the product block, or not sending the email altogether.
For example, if your product block uses Next Best Purchase but doesn't have enough data to generate recommendations for a customer, you could set Best Sellers as a fallback. This ensures customers will see popular products in their email when personalized recommendations aren't available.
Each product block can have up to two fallback options, with the last one being either Do Not Show Product Block or Do Not Send Email.
Configuration limitations
When configuring your recommendation rules, be aware of several limits designed to prevent overly complex setups.
Unique attribute values
For recommendation rules using Best Sellers or New Arrivals, you have the option to base the recommended products on one of the following:
- sitewide.
- category of input product(s).
- a specific attribute that you can define.
If you select Based on: other product attributes, a section named Find related to input product(s) based on appears, where you can select product attributes.
Some attributes may not be available to select from, based on how varied that attribute is in your catalog.
These product attributes aren’t available if there are more than 100,000 unique values of the attribute. Consider how many products would appear if each one had a unique attribute value.
This restriction prevents filtering across a large portion of your catalog, which could degrade performance. We recommend keeping your filters as specific as possible to ensure recommendations are better tailored to each customer.
The following product attributes are always available, regardless of how many unique values exist in your catalog:
- category
- brand
- gender
- division
- subcategory
- p_color
- breadcrumb_0
- breadcrumb_1
- breadcrumb_2
Example 1: Attributes with five unique values
Your catalog has 100,000 products.
The Rating field has five possible values: 1-5, so its unique count is five.
Rating is an allowed attribute because the five unique values are below the 100,000 limit.
Example 2: Attributes with many unique values
Your catalog has 10,000 products.
New products are added to your catalog regularly, creating thousands of unique values for the Created attribute across your catalog. In this example, you have 3,000 unique values for the Created attribute.
The value of 3,000 is below the 100,000 limit. Created is allowed because it doesn't exceed the 100,000 threshold.
Example 3: ID
Your catalog has 2 million products.
Because each product has a unique ID, there are 2 million products with unique values for this attribute.
The ID's value of 2 million exceeds the 100,000 limit, so you would be restricted.
Description contains
When building a product recommendation rule filter using the Description attribute, you must use is/is not criteria.
The Contains/Does not contain options are unavailable for this attribute. When using a description attribute use Is/Is not.
Contains/Does not contain limits
Recommendation rule filters can accommodate up to 12 values per attribute when using the Contains/Does not contain criteria.
There is no limit for attributes using Is/Is not criteria.
If you've already entered more than 12 values and then change the criteria to Contains/Does not contain, only the first 12 values will be kept. Any additional values will be removed.
Additionally, each product block is limited to a maximum of three Contains/Does not contain filters. Because boolean filters and number-based filters do not have Contains/Does not contain parameters, they are not limited by the restrictions.
Cloned campaigns
The above restrictions only apply to newly created and cloned campaigns.
If you try to clone a campaign that violates some of the restrictions, you will see warning messages when adjusting the recommendation rules. You also won’t be able to save and launch the campaign without changing the recommendation rules to follow the guidelines.
Derived attributes
Derived attributes are based on existing attributes. For example, if you have a Color attribute, you could also have a derived "Primary color" attribute with a simple Yes/No value, depending on the item's color.
If you have a use case that would override any of the above restrictions, reach out to your Customer Success Manager. The Bluecore Forward Deployed Engineering team can create derived attributes to help meet your goals with recommendation rules.
Best practices
- Check your email proofs to ensure that there are enough products to satisfy your applied criteria.
- Consult with your Customer Success Manager about recommendation rule design to ensure your blocks aren't overly restrictive, which could increase the chance of halting.
- We recommend adding a fallback to each rule to ensure customers receive product recommendations in all situations.
- Adjust template-specific or global settings for any dynamic product blocks using Product Settings. By configuring global product exclusions on the settings page, you can exclude products from all campaigns, except when customers have previously interacted with them.