Release Summary 25.21 | Oct 16, 2025
The following key features and improvements, along with bug fixes, have been released in Algonomy CXP products in the release version 25.21.
Recommend
Introduced Regionalized Primary Categories
The recommendation engine now supports region-specific primary categories, enabling more accurate and localized product recommendations. This enhancement ensures that category hierarchies reflect the structure defined for each region, improving relevance and alignment with regional catalogs.
This functionality is supported when using a product attribute to define the primary categories. We expect the primary category data to be provided as an attribute that includes both global value and regionalized values. When a regionalized value is not available for a given region, the system will automatically fall back to the global value.
Jira: ENG-30996
Recs Test Drive: Email Recommendation Images Now Open in New Tab
When previewing email recommendations in Recs Test Drive, clicking on a recommendation image now opens the link in a new browser tab. This prevents users from losing their input data while exploring product previews.
Jira: ENG-31111
Recommend Segment Report: View and Download Data Across All Regions
The Recommend Segment Report now supports viewing and downloading data for all regions in a single action. This enhancement eliminates the need for manual, region-by-region selection, saving time and ensuring consistency in analysis.
The updated report includes a multi-select option for flexibility. When no region is selected, all regions are included by default. The visualization shows aggregated data for selected regions, while the table displays detailed region-level data with an added Region column. This provides a clearer and more efficient view of recommendation performance across multiple regions.
Jira: ENG-31071
Find
Find Reports: Enhanced Zero Search Terms Trend Analysis
The Find Reports have been enhanced to display the trend of zero-result queries over time, helping merchandisers measure improvements as the system learns and reduces such occurrences. This update makes it easier to identify how search performance evolves and to focus on improving query handling.
A new “Trend for Zero Search Terms” tab in the Find Search Terms Report shows both the count and percentage of queries returning zero results (without facets) in a clear, date-based line graph. Users can toggle between the number or percentage of queries and select custom date ranges, gaining deeper insight into how effectively search relevance improves over time.
Find MVT Reports: Added Findability and CTR Metrics
The Find Configuration MVT Report now includes Findability and Click-Through Rate (CTR) as new metrics, enabling optimization managers to compare and refine search configurations based on user engagement.
Jira: ENG-31100
Ensemble AI
View Seed Products that Generated Outfits
The Ensemble AI interface now lets merchandisers easily identify seed products that have generated outfits, improving visibility into how ensembles perform across styles and categories. This helps assess model effectiveness and refine strategies based on actual outfit generation.
Users can toggle between potential and actual seed products to view only those that produced outfits. The updated view also displays the total count of seed products and outfits at both style and category levels, providing a clearer picture of overall performance across styles.
Jira: ENG-29975
Introduced Ability to Add Cross-Category Products to Freeform Strategies
The Ensemble AI experience now allows merchandisers to add individual products from different categories while using Freeform mode. This gives more flexibility to experiment with product combinations and generate ensembles that match diverse themes or styles.
Users can now include specific products alongside seed categories to generate ensembles for a defined number of parts. The system intelligently combines the selected products with others, even beyond the chosen categories, enabling broader and more creative outfit generation for testing and optimization.
Jira: ENG-31170
Engage
Support for Multiple Contents per Placement
Engage now supports returning multiple contents within a single placement, making content management simpler and more efficient. This enhancement allows optimization managers to handle fewer placements while delivering more personalized experiences.
A new Multiple Contents layout type has been introduced in the UI, alongside existing Recommend and Content options. It includes similar placeholders, content qualification settings, and promo variable options. Users can set minimum and maximum content limits for each placement, ensuring only qualified content is displayed. Content is automatically deduplicated across campaigns, maintaining relevance and variety.
Social Proof
Support for Aggregate Metrics at Product Level
Social Proof now supports aggregating product metrics across multiple SKUs that share a common attribute (for example, color variants represented as separate product IDs). This enhancement ensures consistent social proof messaging across all related product variants. Both the Social Proof Metrics API and the Social Proof Output Response API now support this feature through a new parameter, requireRolledUpAggregation.
When set to true, metrics are rolled up based on the short name deduping attribute, allowing merchandisers to display unified social proof counts for variants of the same product. This provides a more cohesive shopper experience and improves message consistency across product displays.
Jira: ENG-27998
Other Feature Enhancements
The following feature enhancements and upgrades have been made in the release version 25.21.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Data Science Workbench (DSW): Hive Option Removed from UI |
The Hive radio button has been removed from the DSW Data Source tab following the deprecation of Impala from the DSW stack. The interface now automatically displays the Hue UI within an iframe containing the Hive Query Editor when the page loads, ensuring a smoother and more consistent experience for users. |
16-Oct-25 |
|
|
Find: Find Reports: Identify Discover and Other Requests from Find API Calls |
Find API reporting has been enhanced to differentiate between Discover and other request types based on response styles, providing clearer visibility into how Find APIs are used. The new rollup categorizes Find API calls by response styles such as idAndScores, facetsOnly, productsOnly, and fullResponse. This helps isolate facet-only requests used for Discover from general Find API metrics. Note: Only the rollup has been completed; the UI update is pending. |
16-Oct-25 |
|
|
Find: Related Searches: Default Property Enhancement |
The default suggestion property definition has been updated to include the relatedSearches property in canonicalPropertyDefinitionCollections.jsonl. This ensures that related search terms are now part of the default configuration, improving consistency and ease of use. The updated property definition has been added to a metastore snapshot and successfully tested to confirm proper functionality in related searches. |
16-Oct-25 |
|
|
Find: Find Maintenance Script: Multi-DC Flexibility Enhancement |
The Find maintenance script for deployer alerts now supports data center–wise flexibility, allowing the use of multiple DC hosts. This enhancement applies to the job type “find_deployer_status_report”, ensuring better scalability and monitoring across distributed environments. |
16-Oct-25 |
|
|
Social Proof: Aggregated Metrics Using Short Name Deduping Attribute |
The Social Proof Event Metrics API now supports aggregating metrics based on a short name deduping attribute, enabling counts to be shown at the Key Item level instead of the Key Item Color (KIC) level. This enhancement ensures consistent social proof messaging across product variations. When the requireRolledUpAggregation parameter is set to true, the API aggregates data using the defined deduping attribute (e.g., collection). If not passed, metrics remain at the individual product level. This update improves flexibility in reporting and provides more accurate, product-level insights. |
16-Oct-25 |
|
|
Social Proof: Message Handling and Logging Enhancements |
The Social Proof API has been enhanced to ensure reliability when optimization is enabled. If the inference API fails, the system now automatically falls back to the standard message display logic to maintain continuity. Additionally, both rec_actions and input_actions are now consistently logged in the Cassandra Engage Variation Events table, regardless of whether optimization is active. |
16-Oct-25 |
|
|
Enterprise Dashboard: Content Catalog: Visual Tag Highlights for Inclusion and Exclusion |
Tags in the Content Catalog and content selector widget now visually indicate their usage type for easier identification. Tags used as inclusions are highlighted in green, while those set as exclusions appear in red. |
16-Oct-25 |
|
|
Recommend: DSW: Impala Removed from Stack and UI |
Impala has been fully removed from the Data Science Workbench (DSW) stack as it is no longer in use. All related configurations, references, and code dependencies have been eliminated to streamline the system. In addition, the Impala option has been removed from the Hue UI, ensuring a cleaner interface and preventing confusion for users working with DSW data sources. |
16-Oct-25 |
|
|
Science: Transparent Images: Storage Migrated to Cloudflare |
The storage location for transparent background images has been moved to Cloudflare for improved reliability and performance. The existing directory structure has been retained to ensure compatibility with current configurations. Additionally, the URL template in the site configuration has been updated to reflect the new Cloudflare destination, ensuring seamless access and delivery of transparent images across environments. |
16-Oct-25 |
|
|
Chatbot: Persistent Session Across Site Navigation |
The Ecommerce Sales Assistant Chatbot now maintains an active session as shoppers browse across pages, allowing them to continue conversations without restarting. This ensures a seamless, continuous interaction even when moving between product or category pages. Each chat session is tracked using a unique session ID passed from the client side. The chatbot retrieves the previous conversation history whenever the same session ID is used, maintaining continuity for up to 30 minutes of inactivity. |
16-Oct-25 |
|
|
Chatbot: Hot Swapping for Product Catalog Updates |
The Ecommerce Sales Assistant Chatbot now supports hot swapping of full product catalog updates for smaller catalogs (up to 150,000 products). This allows one catalog to stay live in production while the other is updated in the background. A scheduled job updates the catalog daily, automatically swapping the refreshed version into production. This ensures faster updates, better synchronization with Ensemble AI, and uninterrupted chatbot performance. |
16-Oct-25 |
|
|
Platform: Feed: Handling Missing Product Attributes File |
The Feed to Streaming Connector now gracefully handles scenarios where the product attributes file is missing during ingestion. This ensures uninterrupted processing and prevents failures caused by incomplete feeds. |
16-Oct-25 |
|
|
Platform: Feed: Support for Custom Attributes in FlatFileDeltaFeed |
The Feed to Streaming Connector now supports sending custom attributes through the FlatFileDeltaFeed, enabling greater flexibility in handling product data updates. This enhancement allows additional product-specific attributes to be included seamlessly during delta feed processing. |
16-Oct-25 |
Bug and Support Fixes
The following issues have been fixed in the release version 25.21.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Find: Hybrid Search: Suggestion Component Bug Fix |
A bug causing the suggestion component to break under Hybrid Search has been fixed. The issue occurred because the component was using the Hybrid Search query parser instead of the eDisMax parser. It now correctly uses the eDisMax parser, ensuring suggestions work properly for related search features. |
16-Oct-25 |
|
|
Social Proof: Session ID Missing in Messaging API Calls |
A bug was identified where the session ID was not being passed in Social Proof Messaging API calls when optimization was enabled. While other APIs (such as targeting and track experience) included the session ID correctly, this one failed due to how the URL was being constructed in Client.js. The issue has been resolved. The SP Messaging API now correctly includes the session ID when optimization is enabled, ensuring proper tracking and consistency across all API calls. |
16-Oct-25 |
|
|
Ensemble AI: View Ensembles Option Disabled for Live Styles |
A bug caused the “View Ensembles” option to appear disabled for live styles in production, even though ensembles were already generated. The option only became active after re-saving the style. This issue has now been resolved. The “View Ensembles” button is enabled by default, allowing users to directly access and view ensembles generated for all live styles without needing to re-save them. |
16-Oct-25 |
|
|
Data Engineering, Social Message: Product Event Metrics: Rollup Table Sync Issue Fixed |
The rollup table was previously showing data from only one child product instead of aggregating all related products. This issue is now resolved. The table correctly sums event metrics such as views, add-to-cart, and purchases across all child products for accurate reporting. |
16-Oct-25 |
|
|
Recommend: Domain Redirect Security Fix: CT Parameter Validation Implemented |
A vulnerability was found where the ct parameter in redirect URLs allowed unvalidated external links, leading to potential misuse and domain flagging. The issue is now resolved. Redirects are validated against whitelisted domains, and any suspicious URLs are blocked across both Engage and Recommend modules. |
16-Oct-25 |
|
|
Recommend: Do Not Recommend Recently Viewed Items Rule Fixed |
A bug was causing the “Do Not Recommend Recently Viewed Items” rule to fail, allowing previously viewed products to appear in recommendations across placements like Home Page and Mini Bag. This issue has now been resolved. The rule correctly filters out recently viewed items, ensuring recommendations exclude products that shoppers have already seen. |
21-Oct-25 |