🚀 Release 2024-42

Release date: June 20, 2024

TL;DR : Build Dynamic Categories on the fly, debug group rules via Ranking Diagnostics, and apply numeric attribute operations to Recommendations widgets. We fine-tuned French and German search models and fixed bugs to enhance performance.

🎉 Feature Updates

Create Your Own Dynamic Categories

We’re excited to announce the launch of Dynamic Categories!

What’s new?

Dynamic Categories are user-generated categories that integrate into your product catalog. They update dynamically based on the rules that define them, saving you time and effort.

Categories can be created based on some pre-defined conditions or by selecting products manually. You can further customize the category’s products by creating merchandising rules, just like for any other native category.

Why is this valuable?

  • Build Versatile Categories: You can create categories for seasonal campaigns to dedicated brand collaboration products; the possibilities are endless!

  • Real-Time Updates: As soon as a new product meets the specified conditions for a Dynamic Category, it is instantly added to the assortment. This ensures that the category is always up-to-date with the latest offerings, saving you time.

  • Improved Customer Experience: Create Merchandising rules on Dynamic Categories to adapt the categories on the fly based on your users’ searches and interactions. Dedicated categories for product assortments make it easier for your shoppers to find them.

Visit the documentation article to learn more.


1. Debug Group Ranking with Ranking Diagnostics

Good news for Group Merchandising customers! We are happy to share that Ranking Diagnostics now supports group rules.

What improved?

  • Using the new Group-by option, you can now view grouped recall in Ranking Diagnostics.

  • This new capability allows you to understand the ranking of groups and products within the groups.

  • The group cards display the product-level merchandising operations, such as boost, bury, sequential lock, etc. This makes your debugging workflow more intuitive.

Visit the documentation article to learn more.

2. Numeric Attribute Operations Support for Recommendations

New Numeric Attribute operations for Search and Category rules were launched in a previous release. We are happy to share that these operations are now supported for Recommendations widgets as well.

What improved?

The below operations are now supported for Recommendations widget rules.

  • "Exclude" and "Include Only" operations on Numeric Attributes.

  • Operations on ranges for Numeric Attribute values (in between, equal to, lower than, higher than). For instance:

    • Exclude all products with “price” lower than $10.
    • Exclude products where “inventory” is below 5 units.

Visit the documentation article to learn more.

3. Enhanced German and French Search Models

We have fine-tuned our German and French search models in an effort to enhance multilanguage search ranking.

What improved?

The updated model can now better identify product types for the specific verticals listed below:

  • Healthcare
  • Bike parts
  • Sporting goods, watches, and camping
  • Toys, gifts, and hobbies
  • Motorbike and automotive
  • Musical instruments

This update will improve search ranking for some queries in these verticals. Relevant products will be shown higher up in the search results. Please note that the changes will be live starting tomorrow.


Learn about Discovery Releases with Bloomreach Academy

Check the Discovery Mini Masterclass: Latest Releases course to explore the latest releases and features of Discovery through expert-recorded self-paced video sessions.

🐛Bug fixes

1. Improved Accuracy of Reporting of Visits and Browse Sessions for Relevance by Segment (RBS) and Real-time Segments (RTS)

We've corrected how we report visits and browse sessions on your site, ensuring a more accurate picture of customer behavior. This is to support the implementation of Relevance by Segment (RBS) and Real-time Segments (RTS).

What was the issue?

Earlier, a technical issue caused visits and browse sessions to be split incorrectly when segment information changed during a customer’s visit.

The below feature pixel parameters contain segment information. The visit/session would split whenever the values of the below parameters changed:

  • RBS: customer_tier, customer_country, customer_geo, customer_profile
  • RTS: cdp_segments

Visits and sessions breaking unexpectedly impacted analytics attribution. It inflated visit/session numbers and degraded reporting accuracy and search learning.

What improved?

We’ve addressed the logic behind breaking visits on segment change. This leads to more accurate attribution data and slightly enhanced search learnings.

We have tested this change comprehensively, comparing outputs before and after the change for specific merchants.

Expected Results

  1. Accurate visit reporting: This aligns the number of visits to your expectations and ensures a closer match to other analytics systems, such as Adobe Analytics or Google Analytics.

  2. Improved revenue attribution: The fix ensures more accurate revenue and conversion attribution, an increase in conversion rate and revenue per visit, and a reduced number of visits. Post-release, expect the reported visits to decrease by 10-20% starting the next day, sessions to decrease by 5-15%, and conversion rates and revenue per visit to increase accordingly. Note that there is no actual change in the number of visits to your site, just how we report them.

  3. Enhanced search results: The update enhances search learning and improves ranking algorithms to serve personalized results.

Who does this impact?

Relevance by Segment (RBS) and Real-time Segments (RTS) customers. There is no impact if you didn’t integrate these features.

No action is required on your end. You may reach out to your DXM and discuss if you want us to reconcile data with your external analytics tools.

2. Insights reports improvements

Library Reports have been improved to populate the data according to the selected date range. Earlier, data wasn't filtered correctly by date.


Subscribe to Discovery Release Email Notifications

Subscribe to Bloomreach Discovery Release email notifications by filling out this form.