Release Summary 25.04 | Feb 20, 2025
The following key features and improvements, along with bug fixes, have been released in Algonomy CXP products in the release version 25.04.
Social Proof
Social Proof API Call Count Reporting
A new Social Proof API Report has been introduced, allowing merchandisers to track API usage over time. This report provides insights into API call trends, categorized by methods such as Metrics API and Messages API, with aggregated and de-aggregated views.
The report is enabled only when Social Proof API service is turned on. Users can view API call counts by date range, analyze trends via ThoughtSpot charts, and download detailed reports including Site ID, Site Name, Method, Date, Total Call Count, and Region.
Jira: ENG-29398
Social Proof Optimization - Logging Key Parameters
To enhance Social Proof Optimization, key parameters are now logged to support better decision-making and A/B testing. These include exploration probability, message type & interval, model ID, and the output metric used for optimization.
The system now logs the highest probability score for the selected message type & interval, captures the actual message shown, and records the model ID for tracking version changes. Additionally, the output metric used in optimization is logged, ensuring better visibility into performance.
Jira: ENG-29499
Ensemble AI
Ensemble AI - View Tracking for Styles & Ensembles
Ensemble AI now logs view and engagement data for styles and outfits displayed on item pages and other placements. This enables tracking of click-through rates, outfit interactions, and view-to-conversion metrics to analyze how users engage with ensembles.
The system records outfit views, associated styles, user IDs, placement details, and experiences, ensuring precise tracking similar to recommendation views. These insights help merchandisers assess performance, optimize styling strategies, and enhance engagement with dynamic outfits.
Jira: ENG-28550
Enterprise Dashboard
User History Context in New Arrivals (Recency) Model
The New Arrivals (Recency) model in Configurable Strategies now supports user history-based personalization, allowing optimization managers to filter new products based on a user’s brand or category affinity.
When selecting the model, users can enable Category Affinity or Brand Affinity as a Personalization Seed, ensuring that recommendations align with user preferences. The strategy preview now reflects the selected option in the breadcrumb, and configurations are saved accurately.
Jira: ENG-29726
Delete All Expired Content in Content Catalog
A new "Delete all Expired Content" button has been added to the Content Catalog, allowing digital optimization managers to quickly remove expired content as the catalog grows.
When clicked, an alert displays the number of expired contents that will be deleted, ensuring transparency before proceeding.
Jira: ENG-29728
Other Feature Enhancements
The following feature enhancements and upgrades have been made in the release version 25.04.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Enterprise Dashboard: Disable Unavailable Models in Configurable Strategies |
Configurable Strategies now only display enabled models for selection, reducing confusion and ensuring optimization managers focus on applicable models for their site. Models that are not enabled appear greyed out, with a tooltip explaining the reason and steps to enable them. |
20-Feb-25 |
|
|
Tier 4 & 5 Options for Category Recommendations |
Configurable Strategies now support Tier 4 and Tier 5 category selections for Category Recommendations, providing greater control over category-based recommendations. |
20-Feb-25 |
|
|
Ensemble AI: Ensemble AI - Display Seed Product Details at the Style Level |
Shows a list of products based on applied filters for each category, displaying only those that qualify as seed products. This ensures that users can view only the potential candidates for seed products. |
|
|
|
Find: Reranking in Hybrid Search to Support Boost & Bury Configurations |
Hybrid search results now respect boost and de-boost rules configured via the dashboard, allowing better control over search rankings. Categories, brands, products, and attributes with positive boosts are elevated in the results, while negative boosts are de-prioritized accordingly. |
20-Feb-25 |
|
|
Find: Datadog Monitoring for Find Data Publisher |
We have implemented Datadog monitoring and alerting for Find Data Publisher jobs running on Azkaban. Two new metrics, find.datapublisher.success and find.datapublisher.failed, track job execution status for query writer, query vector, catalog vector, and related searches. A Datadog dashboard has been created to monitor these jobs, and alerts are configured to notify failures in real time. |
20-Feb-25 |
|
|
Find: Re-ranking Hybrid Search Results Based on Catalog & Personalization Settings |
Hybrid search results are now re-ranked based on catalog and personalization settings configured in the search engine. This enhancement improves search result relevancy and ensures alignment with merchandising strategies. The update applies positive boosts for catalog attributes and personalization settings, enhancing results based on affinity and delay settings. |
20-Feb-25 |
|
|
Recommend: Click Tracking for Boosted Products in Avro Logs |
To improve Recommendation Boosting rule analysis, clicks on boosted products are now logged in Avro logs along with the applied boost rule. Now, whenever a boosted product is clicked within a placement, the system logs both the click event and the specific boost rule applied, enabling more precise performance tracking and reporting. |
20-Feb-25 |
|
|
UPS: Filtering Invalid Users in User Linking Feed Processing |
The User Linking feed now filters out invalid empty users to prevent incorrect user mappings. Previously, empty user entries were treated as valid, leading to unintended merging of user linkings. With this update, empty user strings and mappings with fewer than two users are ignored during processing for both Explicit User Linking Feed and Grouped User Linking Feed. All existing BuildFtp, BatchWriter, and UpsBatchSparkJob functionalities continue to operate as expected. |
20-Feb-25 |
Bug and Support Fixes
The following issues have been fixed in the release version 25.04.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Find: Escape Character Handling for Double Quotes in Query Tags |
We have fixed an issue where query understanding failed due to double quotes in query tag values. Now, search service properly escapes double quotes, preventing syntax errors in Solr processing. |
20-Feb-25 |
|
|
Find: Package Shown Even If Inner Products Are Missing in FIND |
A bug was identified where package products were being returned in FIND search results even when all inner products were unavailable. This issue led to incorrect product visibility for customers. The fix ensures that packages are only returned if at least one inner product exists in the catalog. |
20-Feb-25 |
|
|
Find: Find Publish Button Now Publishes to a Single Environment |
The Find Merchandising and Visual Merchandising rules' Publish button no longer displays multiple environment options. Since publishing is only possible in one environment, the button has been updated to function like the Search Term Expansion button, ensuring a consistent and simplified user experience. |
20-Feb-25 |
|
|
Recommend: Configurable Strategies - No Recommendations for Non-Recommendable Seed Products |
The issue with Configurable Strategies returning a non-recommendable seed product without recommendations has been fixed. Now, if the seed product is unavailable for purchase, it will not be included in recommendations when the "Include the seed product with the product recommendations" option is selected. |
20-Feb-25 |
|
|
Recommend: Default AND Value in Action Condition Expression for Rich Rules |
The default value for the AND operator in action condition expressions for rich rules in Configurable Strategies has been implemented. Restriction and Boost Rules now function correctly with both AND/OR conditions. |
20-Feb-25 |
|
|
Data Engineering: /trackingevent API No Longer Fails Without SpOptimisation Parameters |
The /trackingevent API now handles missing SpOptimisation-related parameters without failure. Default values are logged when these parameters are not provided, ensuring the API successfully records tracking data in Cassandra. |
20-Feb-25 |
|
|
Ensemble AI, Science: Region-Specific Ensembles Now Generated in Ensemble AI |
Ensemble AI was not generating region-specific ensembles when region as seed was enabled for styles like Cycling Accessories. This issue has now been resolved, and region-specific ensembles are correctly generated. |
20-Feb-25 |
|
|
Social Proof : Social Proof Output Response API Now Handles Hidden Messages Correctly |
The Social Proof Output Response API was incorrectly returning a blank response when all message types were hidden. Instead of showing an empty experience and variation, the API now correctly retains all details except the final message. |
20-Feb-25 |
|
|
Find: Removal of Unnecessary Package Inner Product Modification in Off-Menu Processing |
An issue in the package parser was causing exceptions due to modification of package inner products in off-menu scenarios. This unnecessary modification led to occasional failures in the Find API under high load. The code has been removed, ensuring stability while maintaining expected functionality. |
20-Feb-25 |
|
|
Language Service Handling When Cluster Pool is Missing |
The language service was failing and restarting when no cluster pool was mapped to a site. Now, the language consumer gracefully handles the exception and skips the event if no Solr cluster is found for the given site. |
20-Feb-25 |
|
|
Data Reporting, Enterprise Dashboard: Year-over-Year Sales Report Display Issue |
The Sales Report was incorrectly displaying only January data for the most recent year in a Year-over-Year (YoY) comparison. Now, the report correctly considers the full year for each selected year. |
20-Feb-25 |
|
|
Find, Streaming Catalog: WoC Data Deletion and Boost Rule Not Applying |
An issue caused WoC data to be deleted from the Solr collection due to an error in the Enrichment code during service restarts. Additionally, a boost rule applied for a specific search term was not reflecting expected score changes. Both issues have now been resolved. |
20-Feb-25 |
|
|
Enterprise Dashboard: Tag Manager Not Displaying Associated Content |
The Tag Manager was not showing content under selected categories due to an API parameter discrepancy. The showExpired parameter in the API call has been updated to showExpired=all, ensuring that content is correctly displayed in the Tag Manager. The issue has now been resolved. |
20-Feb-25 |
|
|
Data Engineering: API Call Rollup Failure |
API Call rollup was failing due to missing viewGuid in certain API logs after the 25.03 release. This issue, which impacted Find API Call reporting, has now been resolved, ensuring successful rollups. |
20-Feb-25 |
|
|
Recommend: Long Description Attribute Not Returning in API Calls |
Long Description Attribute Not Returning in API Calls The issue where the Long_Description attribute was not being returned in API calls when no region ID (rid) was specified has been resolved. The default fallback mechanism is now correctly set to English, ensuring that the attribute is returned consistently, whether or not a region ID is included in the request. |
20-Feb-25 |