Release date: March 6, 2024

TL;DR : This release features the BETA launch of Events Management, new capabilities for LLM-based precision, data debugging via API, and Similar Products Recommendations for French and German sites. We have also introduced updates to Algo Weight Customization, IAM User Management, and Usage app for a seamless user experience.

🎉 Feature Updates

1. Streamline Pixel Integration via Events Management

We are thrilled to announce the BETA launch of our latest feature: Events Management! This feature facilitates the pixel integration process with real-time insights, provides finer control over the event tracking on your site, and enables data-backed debugging.

What’s New?

Events Management empowers you to instantly gauge whether your pixel implementation works as expected. It comes with:

Status module that serves as your event health dashboard.
Events Stream module that serves as a live feed of tracked events on your site.

Events Management also features two targeted operational modes — Integration and Live. We have developed these tools to support you throughout different stages of the pixel integration lifecycle. Here's a quick demo video.

Why is this valuable?

With Events Management application, you can expect faster integration times and better data visibility without the need to contact Bloomreach Support.
Debugging live site event data in real-time brings transparency to data quality and improves data quality. This eventually contributes to better performance of Discovery features.

How can I access this feature?

You must have the 'Events Management User - Read Only' role to access the app.
Navigate to 'Setup > Events management' in the Bloomreach dashboard and see the feature in action!

Feedback

As Events Management is in the BETA phase, we welcome and encourage your feedback to refine capabilities for future iterations of this application.

Explore the documentation to learn more.

2. LLM-Based Precision Mode is now Generally Available

We are excited to share that LLM-based precision is now generally available for all English catalogs.

What’s new?

Ability to enable the precision mode by passing the below parameter in the front-end API call: query.precision=llm_based_precision
AB testing is now supported. Pass the above-mentioned parameter to configure the test and evaluate the fine-tuned recall.
This feature is free to use until July 31, 2024.

Why is this valuable?

With enhanced LLM-based precision capabilities, you can utilize the latest generative AI technology for better search recall precision. Visit the documentation to learn more.

3. Product Search and Category API: Preview rules active in future with new parameter

We are excited to share that we are adding a new parameter—merchandising.ts— to the Search and Category API. The parameter lets you preview merchandising rules before they become active at a future date.

What’s new?

Preview how product pages and merchandising rules will look at a future date, taking into account changes that will occur over time.
When you include the parameter in the Search API request, the front end uses the timestamp to fetch the applicable merchandising rules.
If the parameter is not sent, the request fetches the currently active merchandising rules by default.

Why is this valuable?

With better visibility into future customizations, you can effectively plan merchandising campaigns way in advance and conquer peak periods like Black Friday or Christmas.

Visit the documentation to learn more.

4. Ranking Diagnostics on the APIs: Fetch debugging data directly via the APIs

We’re happy to share that Ranking Diagnostics are now available on the APIs. With this update, you can directly fetch debugging data for product ranking, recall, and merchandising in your API requests.

What’s new?

Fetch all the debug info directly on the Search and Category APIs using a new parameter &br_diagnostics=all.
Fetch specific debug data on the API by passing different values to the new parameter &br_diagnostics=merchandising, &br_diagnostics=recall OR &br_diagnostics=ranking.

Why is this valuable?

These insights enable you to understand the algorithm better or troubleshoot ranking/recall issues of products.

Already accessible through the dashboard, this Ranking Diagnostics data is now also available on the API for developers, support teams, and other personnel to use.

Visit the documentation to learn more.

5. Similar Products Recommendations now in French and German

Good news for Similar Products Recommendations customers! Similar Products will now power French and German catalogs.

What’s new?

Our cutting-edge algorithm understands the attributes associated with products and produces relevant recommendations for French and German language input.

Why is this valuable?

This feature is useful for customers seeking multilingual support for Similar Products recommendations. Stay tuned, as we will soon launch this capability for other non-English languages as well!

⚡Improvements

1. Algo Weight Customization: New option to apply global rules

Algo Weight Customization features a new option called “Use global settings.” This option ensures that the global Custom ranking algorithm rules are applied at the local levels automatically.

What improved?

Now, the algorithm customization rules created at the global level will apply to the lower levels by default. If you wish to override the global rule, you can simply create a customization rule at the individual category/query level.
Earlier, the settings would default to the option “Use default ranking.”

2. Identity and Access Management (IAM): Multi-Organization User Management

We have improved the way users can be managed via IAM. Users can now belong to multiple Discovery organizations simultaneously.

What improved?

Ability to link users to multiple organizations in Discovery IAM.
Users can register once and associate with several organizations. They no longer need to manage multiple email accounts to access different organizations.
Users can easily move between accounts and organizations using the Site selection dropdown. This facilitates seamless collaboration.

If you have users who work with multiple organizations, this allows greater flexibility and control to add/remove users for temporary projects.

Earlier, users were limited to a single organization and needed separate emails for each organization.

3. Usage Reporting Improvements

We have added the below enhancements to the Usage Dashboard:

Theme update to ensure interface consistency.
Group by “All” is now “(none)”
Contrasting colors for the bar chart.
Request Types are now visible in alphabetical sort.
Date Range picker is now in UTC time standard, the same as the chart and report.

4. Sleek New Look for Discovery Dashboard Interface

We continuously strive to make your experience with the Discovery dashboard seamless and intuitive. Here are recently introduced UI updates that elevate the dashboard’s look and feel:

Enhanced styling of the main navigation.

Refined the UI copy to make it clear and easy to understand.

Enhanced styling of all buttons (except menu buttons) throughout the Merchandising modules. You may have noticed that some buttons have been changed from blue to yellow. This ensures consistency with other Bloomreach apps.

🐛Bug fixes

Autosuggest API: We have improved the handling of prefixes with accented characters. Earlier, a request with accented input returned a 5xx error response.
Merchandising audiences: In the Location dimension, the suggestions dropdown would not retrieve the country entered. We have resolved this issue. You'll now be able to search and select the country from the suggestions dropdown.

📘
Tips to Search: Remember to search by following the enforced template. Type in the name of the country, followed by the name of the state and then the city. For example: united states,new york,sodus

Insights: We improved how products are joined in reporting for specific multi-site customer integration cases. Before, some product thumbnails and metadata did not load properly.
Staging API: The fields with spaces were throwing 500 Internal Server Error. The error is now fixed.