Release Summary 25.03 | Feb 07, 2025
The following key features and improvements, along with bug fixes, have been released in Algonomy CXP products in the release version 25.03.
Ensemble AI
Merchandising Report for Ensemble AI
A new Merchandising Report is now available for Ensemble AI, allowing merchandisers to track product performance at the product, category, and brand levels. This report provides key insights into how products perform within ensembles and similar product recommendations.
Key metrics include Attributable Sales, CTR, Clicks, Items from Ensembles, Overall Items, Page Views, and Sales. Additionally, product views now indicate whether the metric is derived from ensembles, similar products, or both, helping merchandisers evaluate the value generated from each.
Jira: ENG-29247
MVT Testing for Ensemble AI On/Off
A new MVT test type has been introduced for Ensemble AI, allowing merchandisers to run an On/Off A/B test to measure its impact. The test leverages the "enable ensemble ai" site configuration, where treatment enables it and control keeps it disabled, ensuring a clear performance comparison.
Eligibility criteria ensure that only visits where products with ensembles are displayed are considered in both control and treatment groups. If no ensembles are shown in a visit, that visit is marked ineligible. Additionally, standard test setup options such as test name, traffic percentage, test duration, and concurrent test flag are included. Merchandisers can also create an A/A test to validate traffic distribution and test assignments.
External Product ID and Site Link in Ensemble Review
Merchandisers can now view the external product ID and product site URL while reviewing ensembles, making it easier to identify and navigate to seed products in the catalog. This enhancement applies at both the style level and across styles for seed products.
Additionally, a new icon has been added to the product image, allowing users to open the product catalog page in a new tab for quick reference.
Jira: ENG-29652
Enterprise Dashboard
New Workflow for Creating Engage MVT
A new Engage MVT test type has been introduced, allowing optimization managers to test non-live campaigns before activating them in production. Unlike the previous Engage MVT, which only allowed testing of live campaigns, this update provides flexibility similar to Merchandising Rules tests, enabling users to evaluate campaign impact before going live.
The test creation flow includes defining test details, selecting campaigns, configuring treatments, and setting traffic distribution. Users can enable or disable campaigns within the treatment and optionally add treatment-specific JavaScript. Existing Engage MVT tests will remain accessible, but only new tests will use the updated "ENGAGE_RULE_ON_OFF_V2" type.
Dynamic Experiences Template for Product Comparison Placement
A new Dynamic Experiences template called Comparison Placement is now available, allowing optimization managers to add a product comparison section to their site for easier upselling of similar products. This template enables customization of style elements, such as font family and size, and includes configuration options for region settings, currency symbols, and API keys.
Users can select placements of type Recommend, define a location selector, and view all attributes in a comparison table.
Jira: ENG-29561
Reccommend
AND/OR Logic for Combining Conditions in RichRules
Merchandising rules now support AND/OR logic when combining multiple conditions in Recommendation Restriction (Only Recommend / Do Not Recommend) and Recommendation Boosting rules. This enhancement gives Optimization Managers greater flexibility to filter or boost products with more precise control, ensuring highly targeted recommendations.
Users can toggle between AND and OR logic by clicking a styled, interactive link within the UI. The default logic is set to AND for restriction rules and OR for boosting rules, while multiple values within a single condition (e.g., multiple brands) always use OR logic. Additionally, when multiple attributes are involved, users can select their preferred logical operator.
Find
Find - New Spell check plugin for Hybrid Search
A new spellcheck plugin has been introduced for hybrid search, enhancing accuracy when no text results are available. This plugin replaces the standard spellcheck feature in our search system, providing improved query corrections and handling cases where the regular spellcheck falls short.
Jira: ENG-29667
Announcing google Product Feed integration with Streaming Catalog API
The Catalog Import process now accommodates the Google Product Feed (GPF) format, streamlining onboarding for customers who are already utilizing Google feeds for their product listings. This enhancement facilitates the seamless integration of product data with Google Shopping, reducing setup time. The system processes the Google Product Feed and imports it into the personalization cloud through the Streaming Catalog API.
Jira: PLAT-3628
Data Engineering
Filtering Bot API Calls in Find API Call Count
The Find API call count now excludes bot-generated API calls (flagged as isRobot=true) from billing calculations, ensuring that only valid user-generated requests are counted. Bot calls are separately tracked and reported for improved transparency in API usage.
Jira: ENG-27294
Channel Data Added to Scorecard Session Report
The Scorecard Session Report now includes Channel ID and Channel Name, enabling internal users to analyze sessions by channel and identify potential instrumentation issues. This enhancement provides deeper insights into how different channels impact session performance.
Channel details have also been added to the rollup, and the visualization has been updated to reflect channel-based insights.
Jira: ENG-28859
Other Feature Enhancements
The following feature enhancements and upgrades have been made in the release version 25.03.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Ensemble AI: Merchandising Report for Ensemble AI |
A new Merchandising Report is now available for Ensemble AI, allowing merchandisers to track product performance at the product, category, and brand levels. This report provides key insights into how products perform within ensembles and similar product recommendations. Key metrics include Attributable Sales, CTR, Clicks, Items from Ensembles, Overall Items, Page Views, and Sales. Additionally, product views now indicate whether the metric is derived from ensembles, similar products, or both, helping merchandisers evaluate the value generated from each. |
07-Feb-25 |
|
|
Recommend: rrGuid Added to Recommend Responses |
The rrGuid value is now included in Recommend API responses, providing a stable user identifier for easier session tracking. Unlike rcs, which varies between requests, rrGuid offers consistency for better user management.
The rrGuid is now available in recsForPlacement and p13n_generated.js responses at the same level as rcs. A Site Configuration allows customers to disable rrGuid if needed, but it is included by default for seamless integration. |
07-Feb-25 |
|
|
Enterprise Dashboard: Non-Blocking Strategy Messages API Call |
The Strategy Rules page now loads faster by making the Strategy Messages API call non-blocking. Previously, the page was delayed while waiting for responses for all disabled strategies, causing slow performance. With this update, the API call no longer blocks page loading, improving the user experience and ensuring faster access to strategy configurations. |
07-Feb-25 |
|
|
Find: Low-Touch Activation of Query Understanding for Streaming API Sites
|
Query Understanding can now be enabled without requiring customers to create a new property definition collection and map it to a new snapshot. This update allows Find to activate Query Boost across existing attributes in the current property definition collection, reducing setup complexity.
Once the query tag collection is mapped to an existing active snapshot, the system updates Find index settings, ensuring Query Understanding functions seamlessly. This includes support for boosting, filtering, disabling, and query reformulation. |
07-Feb-25 |
|
|
Find: Override Configurations in Runtime APIs
|
Runtime APIs now allow selective overrides of Query Understanding (QU) and Hybrid Search settings without modifying the full complementary JSON. Customers can pass specific parameters with the hybridSearch prefix in Find API requests to control search behavior dynamically.
|
07-Feb-25 |
|
|
Find: Flexible Scheduling for Query Vector Jobs |
Query vector jobs now support scheduling at daily, hourly, and minute-level intervals, improving flexibility in search query processing from MongoDB. The job dynamically determines the date range using midnight truncation and look-back logic to ensure no records are missed while optimizing data retrieval. |
07-Feb-25 |
|
|
Data Engineering: MVT Reporting for Ensemble AI |
MVT reporting now supports the Ensemble AI test type, enabling accurate tracking and analysis of A/B tests. The report ensures eligibility by considering only visits where products with ensembles were displayed on both control and treatment sides. If no ensembles were shown, the visit is marked ineligible.
|
07-Feb-25 |
|
|
Discover: Raw Counts for Views, Clicks, and Purchases in Find Solr |
The Find Discover API now includes raw counts for views, clicks, and purchases to help merchandisers assess product popularity. These metrics, based on the 'pcs lookback number of days' Site Config, provide deeper insights into product performance. The data is now retrieved and stored in the enrichment service for improved Discover Workbench support. |
07-Feb-25 |
|
|
Engage: Option to View Only Inactive Contents |
The content API now allows filtering to view only inactive contents, including non-recommendable and expired items. This enhancement helps optimization managers easily identify and delete outdated content. The API now supports filtering by recommendable, non-recommendable, active, and expired statuses. |
07-Feb-25 |
|
|
Engage: Improvements to Dynamic Experiences API |
The Dynamic Experiences API now filters experiences based on context and segment, ensuring that only relevant experiences are returned for a given request. This enhancement enables customers to use Dynamic Experiences without client.js, making it compatible with mobile apps. Additionally, a new parameter allows excluding HTML, CSS, and JavaScript from the response, optimizing API usage for server-side implementations. |
07-Feb-25 |
|
|
UPS: Improved User Linking and Segmentation Process in UPS
|
The user linking and segmentation process has been enhanced to improve reliability and monitoring. The system now tracks file records per site, raises alerts for failed file persistence, and provides an API/dashboard for real-time status updates. Key improvements include support for up to 3GB of grouped linking files and 5GB of user-linking files, with failures triggering Slack alerts for manual intervention. |
07-Feb-25 |
Bug and Support Fixes
The following issues have been fixed in the release version 25.03.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Find: Assortment Not Displayed for eShop DK Production |
An issue was identified where certain products, such as "DK_6920526," were not appearing as assortments in eShop DK Production, despite having the correct data in the view-store. This caused the assortmentSection to appear empty when users searched for specific article numbers. |
07-Feb-25 |
|
|
Find: SFI Not Caching Internal IDs of Items |
A bug caused SFI to miss caching internal IDs of items, leading to repeated calls to the identity service and performance slowdowns. The issue has now been fixed, ensuring that internal IDs are properly cached, improving efficiency and reducing latency in SFI operations. |
07-Feb-25 |
|
|
Find: Package Parser Showing Random Behavior |
The Package Parser was failing inconsistently across Solr nodes due to issues with the visibility filter cache in Solr 7, causing syntax errors and zero results at times. This issue has now been fixed, ensuring consistent behavior across all nodes. |
07-Feb-25 |
|
|
Recommend: Incorrect Affinity Score Calculation for Attributes and Brands in Dashboard API |
The /affinityScores API was incorrectly calculating affinity scores for attributes and brands, always using a weight of 100, leading to inaccurate results. This issue only affected the dashboard API, while runtime campaigns continued to function correctly. This issue has now been fixed, ensuring that changes in attribute weights properly reflect in the API response. |
07-Feb-25 |
|
|
Recommend: Description Field Missing in getPlacementsForPageType API |
The getPlacementsForPageType API was not returning the description field for placements, causing issues for clients like Matas who rely on dynamic configurations instead of hardcoding. The response contained empty description fields despite being published. This issue has now been fixed, and the API correctly returns the populated description field for placements based on the specified page type. |
07-Feb-25 |
|
|
Engage: /content API Returning Non-Recommendable Content |
The /content API was incorrectly returning non-recommendable content due to relying on the wrong attribute. Instead of using the recommendable flag, it now correctly references the r_recommend attribute to determine content eligibility. With this fix, the API now only returns recommendable content by default. If showRecommendable=false, it correctly filters and returns only non-recommendable content. |
07-Feb-25 |
|
|
Engage: Engage MVT Setup Issue |
The Engage MVT test setup has been improved to simplify testing content variants against a control. Previously, users had to manually adjust campaign priority, making the process confusing. Now, the test allows specifying control and variant campaigns directly, ensuring a more intuitive setup. With this fix, Engage MVT now automatically publishes all test variants and the control, eliminating the need for manual priority adjustments. |
07-Feb-25 |
|
|
Enterprise Dashboard: XSS Error in Manual Recommendation Rules |
An issue where an XSS error was triggered when saving Manual Recommendation rules with product names containing parentheses has been fixed. The solution ensures that the front-end no longer sends product names in the payload, preventing potential security risks. Now, only the product ID is sent when saving a rule, while the API dynamically retrieves and displays the correct product name in the UI. |
07-Feb-25 |