Release Summary 25.17 | Aug 21, 2025
The following key features and improvements, along with bug fixes, have been released in Algonomy CXP products in the release version 25.17.
Engage
Content Test Drive Enhancements: Improved Attribute Handling, Validation, and UX
Several enhancements have been made to the Content Test Drive to improve usability and accuracy. When displaying images in results, the system now correctly matches image attribute keys by converting them to uppercase to align with the format returned by the Personalize API. Channel selection has also been validated—if left empty, a clear alert prompts users to select a channel before proceeding, avoiding misleading error messages. This validation has been extended to both Content and Recs Test Drives.
Additionally, content links now open in a new tab, preserving the input form, and the “no results” message now refers to “content” instead of “recommendations.” Lastly, the Campaigns tab now correctly displays both rule IDs and names in the appropriate columns.
Jira: ENG-30641
Social Proof
Support for Multiple Prediction API Instances in Model Consumer
The model consumer now supports multiple Prediction API instances, enabling better scalability in production environments. The inference API is configured to work through the Marathon URL, allowing multiple instances to run concurrently and serve requests efficiently.
Jira: ENG-30854
ATC Rate Now Included in Social Proof MVT Reports
To help merchandisers better evaluate the effectiveness of social proof messaging, Add to Cart (ATC) rate has been added as a key metric in MVT reports for Social Proof tests. This metric provides insight into user engagement where Add to Cart is the primary call to action.
The report now includes item page views, add to carts, and ATC rate (calculated as add to carts divided by item page views). These appear after the conversion rate metric, along with lift and confidence values for ATC rate. Lift values are also included for item page views and add to carts, enhancing the ability to compare performance across test variations.
Jira: ENG-30611
Chatbot
Enhanced Chatbot Integration with Click Tracking and Secure API Support
A new backend API is now available to generate click tracking URLs for products recommended within chatbot sessions. This service accepts placements, product IDs, chat ID, and turn index, and returns clean product URLs along with tracking-enabled click and redirect URLs. It also introduces backend support for a new chatbot_page page type, with UI configuration support to follow.
Additionally, the chatbot now uses updated API endpoints with OneLogin authentication for improved security and consistency.
Enterprise Dashboard
Feedback Added to Dashboard Chatbot Responses
Users can now provide quick feedback on chatbot answers using thumbs up or thumbs down icons shown below each response. Positive feedback is instantly submitted, while negative feedback opens a textbox for optional comments to explain what went wrong. Each response allows only one feedback action, and once selected, the choice is final. This enhancement helps improve future chatbot interactions by capturing valuable user insights.
Jira: ENG-30692
Support for InRegion Option in Omni and Offline Top Sellers Models
Configurable Strategies now support the “Use Region as a Seed” option for both Top Sellers Omni Channel and Top Sellers Offline models. This enhancement enables the use of InRegion models to reflect regional shopping behaviors more accurately while improving performance by reducing post-processing filters. It's particularly beneficial for retailers with large regional footprints.
Jira: ENG-30795
Other Feature Enhancements
The following feature enhancements and upgrades have been made in the release version 25.17.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Social Proof: Informational Message Added for Social Proof Optimization |
When optimization is enabled at the variation level in Social Proof, an informational message now clarifies how the system selects messages. It indicates that the engine will explore all message options over a period before selecting the best-performing one for the configured KPI. This helps merchandisers better understand the impact of enabling optimization during message setup. |
21-Aug-25 |
|
|
Social Proof: Optimization of Social Proof Experience API Calls |
Enhanced the Social Proof experience API by using additional parameters to fetch only relevant and qualified experiences. This eliminates the need for separate targeting requests and manual filtering based on context and segment IDs. While the logic remains unchanged, this optimization reduces unnecessary API calls and improves latency. |
21-Aug-25 |
|
|
Mixpanel Integration for Dashboard Analytics |
Mixpanel has been integrated across all key dashboard pages to enable product teams to track and analyze user behavior. Using Autocapture, the system now collects standard interaction events (clicks, inputs, page views), along with key user session properties like user ID, email, site ID and name, and role information. Events are now visible in the Mixpanel dashboard for real-time insights. |
21-Aug-25 |
|
|
Find: Find Event Logging for Bandits |
Logging has been added to the Find events and visits parquet to support Bandits model requirements. This enables capturing of additional search configuration data and facilitates mapping of query-click events to the configurations used at the time. |
21-Aug-25 |
|
|
Find: Automated Job for Generating Related Searches Using LLM |
A new backend job has been implemented to automate the generation of related search terms. This job retrieves search queries from MongoDB, generates relevant suggestions using OpenAI's language model, and ingests the results into the streaming pipeline for downstream consumption. |
21-Aug-25 |
Bug and Support Fixes
The following issues have been fixed in the release version 25.17.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Enterprise Dashboard: Mobile View Optimization for Guided Selling |
The Guided Selling quiz now dynamically adjusts its screen height based on each question’s content. This ensures that the "Next" and "Back" buttons always appear right below the last answer, eliminating excess space and avoiding scroll issues—especially when moving between image-heavy and text-only questions. |
21-Aug-25 |
|
|
Enterprise Dashboard: Fixes to User Affinity Configurations Page |
The User Affinity Configurations page has been updated to improve display accuracy and user experience. Affinity graphs are now hidden if all scores are zero, ensuring clean visualization. Event scores within each graph are properly sorted by total score, including for Browsed Category events. Additionally, accented characters like “Rosé” are now handled correctly. |
21-Aug-25 |
|
|
Recommend: Recs Test Drive - Strategy Selection Bug Fixed When Changing Placements |
A bug in Recs Test Drive has been resolved where switching placements after selecting a strategy caused the strategy to appear selected in the UI but passed a null value in the API request. This led to errors and no recommendations being returned. The issue has now been fixed, and the selected strategy is correctly retained in the request payload when changing placements. |
21-Aug-25 |
|
|
Recommend: Error While Editing Product Conditions in Recommend Restriction Rules |
Resolved an issue where editing a “Do Not Recommend” or “Only Recommend” rule with both brand and product conditions caused an error when removing a product. A new Reset link has been added to each action tab (Only Recommend, Do Not Recommend, Boost) to clear selections cleanly, ensuring the rule can be saved without any residual conditions. |
21-Aug-25 |
|
|
Recommend: Custom Parameters Not Reflected in Recs Test Drive |
An issue was resolved where custom parameters added in Recs Test Drive were not reflected in the HTML request. The fix ensures that custom inputs are now properly incorporated into the request payload. |
21-Aug-25 |
|
|
Guided Selling: Screen 2 Display Issue Resolved |
A bug that caused Screen 1 details to reappear when clicking on Screen 2 during initial quiz setup has been fixed. This occurred only the first time questions and answers were being entered. Now, selecting Screen 2 correctly displays only its corresponding details. |
21-Aug-25 |
|
|
Guided Selling: Mobile Layout Issue with Image + Text Answers Fixed |
An issue where only one image appeared per row on mobile when answer options included both image and text has been resolved. The layout now correctly displays two images per row, improving visual consistency and usability on mobile devices. |
21-Aug-25 |
|
|
Chatbot: Chatbot History Print/Save Now Supports Full Conversation |
An issue where the chatbot’s print/save functionality was truncating the chat history after the first page has been resolved. Users can now save or print the entire conversation, regardless of its length. |
21-Aug-25 |
|
|
Chatbot: Single Feedback Selection Enforced in Chatbot |
An issue where users could click both the thumbs-up and thumbs-down icons on a chatbot response has been resolved. Now, selecting one feedback option disables the other, ensuring only one type of feedback can be submitted per response. |
21-Aug-25 |
|
|
Incorrect Category Handling in Configurable Strategies |
Resolved an issue where selected subcategories (e.g., “Hoodies & Sweatshirts”) were incorrectly replaced by their parent category (“Tops”), resulting in broader product recommendations. This also affected Test Drive results. Category selections now persist correctly, ensuring accurate recommendations and previews. |
21-Aug-25 |
|
|
Category Selection Not Persisting in Configurable Strategy |
Fixed an issue where the selected category “radsport” would revert to its parent category “sportarten” after saving and reopening a configurable strategy. The system now correctly retains and displays the selected category. |
21-Aug-25 |
|
|
Viewed Together Model Not Returning Recommendations for Some Users |
Resolved an issue where the Viewed Together model failed to return recommendations for certain users despite having sufficient browsing history. The root cause was identified and corrected, ensuring consistent behavior across all users in configurable strategies using this model. |
21-Aug-25 |
|
|
Find: Query Understanding Not Applied to 'AND' Brand |
Resolved an issue where QueryUnderstanding logic was not correctly applied to the “AND” brand for an user, causing inaccurate search results. Although the brand was present in the catalog tagger collection, it was not being recognized in query parsing. The backend has been updated to correctly apply QueryUnderstanding to this brand, ensuring accurate tagging and relevant recommendations. |
21-Aug-25 |