Release Summary 26.18 | Sep 03, 2026
The following key features and improvements, along with bug fixes, have been released in ADA Digital Experience Personalization (DXP) in the release version 26.18.
Recommend
Placement Profile Page Now Generally Available
Every placement now has its own Placement Profile page, a single view that brings together everything influencing what shows up there, so you don't have to piece it together across separate screens.
The page includes:
-
Every strategy playing on the placement, including its priority and fallback order
-
All rules targeting the placement, organized by type: Strategies, Merchandising, Filters (Only Recommend and Do Not Recommend), and Boost
-
The placement's page type, layout, and configuration settings
-
Performance data for the placement and for each strategy, including views, click-through rate (CTR), and attributable sales, with trend comparisons over 7, 14, or 30 days
A few additions make day-to-day management faster:
-
Strategy messages can be edited directly from the page, without navigating to the Strategy Configurations screen
-
Rules are grouped by label into collapsible sections, with search and a Production and All Rules toggle to narrow things down quickly
-
The placement list view surfaces the placements that matter most, sorted by views and attributable sales, with search to jump straight to a specific one
The Placement Profile page can be accessed from the Recommend section by clicking into any placement to see its full profile.
Jira: ENG-32956
Ensemble AI
Virtual Try-On Preview for Outfit Review in Ensemble AI
Merchandisers reviewing outfits in Ensemble AI can now preview virtual try-on images directly within the review workflow. Virtual try-on images are generated automatically as part of the ensemble creation process and made available through both the portal API for merchandiser review and the front-end API for the shopper experience. If the virtual try-on job encounters an issue, outfit generation is unaffected, with any failure surfaced as a clear, user-friendly error.
A toggle at each outfit level lets merchandisers switch between the standard outfit view and its virtual try-on rendering, available at the individual outfit level, across all outfits for a given seed, and across styles. The option only appears when virtual try-on is enabled at the site level and is automatically disabled for outfits that do not yet have a virtual try-on image, so merchandisers always know which outfits are ready for preview.
This gives merchandisers a faster, more visual way to validate outfits before they go live, reducing back-and-forth and helping catch styling or model issues earlier in the review process.
Boost Same-Brand Pairings in Ensembles
Merchandisers can now boost products from the same brand when generating ensembles, making it easier to ensure items from a given brand appear together in outfit recommendations. A new "Boost same brand ensembles" option is available at the style level, backed by a new API that lets the dashboard configure this preference per style, applying a 25% boost to the computed score of same-brand product pairs when enabled.
This boost works alongside existing signals rather than overriding them: co-purchase data continues to take priority, and pairs with stronger purchase or view co-occurrence can still rank higher, keeping ensembles grounded in genuine shopper behavior. This gives merchandisers a practical way to reinforce brand representation in outfit recommendations while preserving the relevance built from actual customer signals.
Enterprise Dashboard
Co-occurrence Report Now Limited to Primary Categories
The Co-occurrence Report has been updated to process and display data for primary categories only. Previously, both primary and non-primary categories were included in processing, which led to inconsistencies in the report. Non-primary categories are excluded going forward, and clients should ensure their primary categories are defined according to their intended reporting needs. The report now includes a label indicating that results are available only for primary categories, and the previous checkbox for selecting primary categories has been removed, since this is now the default behavior.
This change improves the accuracy and reliability of the Co-occurrence Report, giving merchandisers cleaner, more consistent insights into category-level product relationships without the confusion introduced by non-primary category data.
Jira: ENG-32830
Social Proof
Enhanced Tracking for Social Proof Messaging
Social Proof messaging now captures additional context with each message served, including region, segment, and the message template used.
With region, segment, and template information captured as part of the tracking data, it becomes possible to analyze social proof performance more precisely, whether by geography, audience segment, or message variant. This lays the groundwork for more targeted reporting and deeper insight into how social proof messaging influences shopper behavior across different regions and segments.
Jira: ENG-32993
Live Preview for Social Proof Badges
Social Proof now supports a live preview for badges, showing exactly how badge will look once published, without needing to enter or embed the client website URL. The preview renders the full badge design, including fonts, colors, borders, shapes, icons, and message text and layout, on a standalone component set against a neutral background.
Any change to the design is reflected instantly in the preview, and repeated updates replace the previous rendering rather than stacking on top of it, a behavior that also applies to the existing social proof messaging preview.
This gives users a faster way to visualize and validate badge designs before they go live, without needing a full website integration to check how a design will appear.
Jira: ENG-32302
Engage/Content Catalog
Regionalized Attribute Support in the Content Update API
The Content Update API now supports region, language, and currency-specific overrides for content attributes such as headlines and calls to action. Content updates submitted through the API are processed and stored with these localized variations, so a single content item can carry different attribute values for different regions, languages, or currencies rather than requiring separate content records for each.
This makes it easier to deliver regionally tailored content, such as localized promotional messaging or currency-specific pricing labels, while managing the content from one place.
Jira: PLAT-4412
Other Feature Enhancements
The following feature enhancements and upgrades have been made in the release version 26.18.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Ensemble AI: Virtual Try-On Generation Insights |
Merchandisers and product stakeholders now have visibility into virtual try-on generation activity across styles. Outfit and virtual try-on counts are captured for every style where the feature is enabled and made available through a reporting layer, making it possible to answer questions such as how many try-ons were generated in a given period and which styles are driving the most engagement. |
03-Sep-26 |
|
|
Enterprise Dashboard: Usage Tracking Added to the Online Help Landing Page |
The Online Help (OLH) now includes usage tracking, capturing which sections and pages users visit across Core Platform, Solutions, Developer Resources, Training Videos, and Product Updates. With this visibility into which documentation topics are used most, the documentation team can prioritize content updates and product decisions based on actual usage patterns rather than assumptions. |
03-Sep-26 |
|
|
Ensemble AI: Image Data Kept in Sync with the Catalog for Ensembles |
The image data used to generate ensembles is now kept in sync with the product catalog, in the same way text-based product data already is. As products are removed or marked non-recommendable, they are automatically excluded, and new products are added as they become available. This ensures ensembles are always generated using current, recommendable products, reducing the risk of outdated or ineligible items appearing in outfit recommendations. |
03-Sep-26 |
|
|
Data Engineering: Coverage Analysis Reporting for Aggregated Metrics |
Coverage analysis reporting is now available for sites that use aggregated metrics across multiple SKUs based on a product attribute, in addition to sites using individual per-SKU metrics. Previously, coverage analysis did not account for this aggregated data, leaving a gap in reporting for such sites.
The report now automatically shows aggregated or non-aggregated metrics based on the site's configuration, giving merchandisers consistent access to coverage analysis regardless of how a site's metrics are structured. |
03-Sep-26 |
|
|
Find: Import Request in the Find Test Drive Workbench |
The Find Test Drive Workbench now includes an Import Request option that lets users paste an existing Find API request URL and have the workbench automatically populate the corresponding fields. This covers request configuration, advanced options, search configuration, hybrid search configuration, and query understanding settings, including embedded search configuration details, saving users from having to enter each parameter manually. This makes it faster to test existing production queries or reproduce customer-reported search scenarios, reducing manual setup time and the chance of configuration errors when working with real-world request URLs. |
03-Sep-26 |
|
|
Enterprise Dashboard: Edit Experience Name During Dynamic Experience Creation |
Users can now edit the experience name while creating a new Dynamic Experience, without needing to complete creation first and rename it afterward. An Edit option opens the same sidebar used on the Landing Page, allowing the name to be updated directly within the creation flow. The updated name is retained as the user continues working and is only saved to the backend when the corresponding variation is saved. This gives users more flexibility to name experiences accurately as they build them, reducing the need for a separate rename step after creation. |
03-Sep-26 |
|
|
Enterprise Dashboard: Faster Navigation Between Placements |
Navigating between placements within the same page type is now quicker. Returning from a placement profile page now lands directly on that page type's tab in the placement list, instead of returning to the All-Placements view and requiring the page type to be re-selected. This reduces the number of clicks needed to move between placements, making it easier for ecommerce managers to review placement configuration and performance across a page type. |
03-Sep-26 |
|
|
Enterprise Dashboard: Streamlined Context Configuration for Manual Recommendation Rules |
The Context section of manual recommendation rules now matches the configuration experience used in Recommend restriction rules. Category and placement selections use a dedicated picker window, showing only the selected items instead of requiring merchandisers to scroll through the full list, and the section now uses the same tab-based layout as other rules to make better use of space. This makes it easier for merchandisers to review and understand manual recommendation rules at a glance, without wading through unrelated configuration details. |
03-Sep-26 |
|
|
Recommend: Improved Accuracy for the Dashboard Chatbot |
The dashboard chatbot's responses have been refined to stay strictly grounded in ADA DXP documentation, release notes, and the context available at the time of the question. When the available information does not fully answer a question, the chatbot now clearly states what it can and cannot confirm rather than guessing, and questions involving specific account data or performance metrics are redirected to the appropriate CSM or CSA contact. This gives users more reliable, traceable answers from the chatbot and reduces the risk of inaccurate or speculative responses. |
03-Sep-26 |
|
|
Find: Expanded Currency and Keyword Recognition for Price Intent in Search |
Price intent in search now recognizes a wider range of ways shoppers express prices. Searches that include "Rs" or "INR" as a prefix or suffix to an amount, in various formats and cases, along with the ₹ symbol, are now correctly understood, so a search like "shirts under Rs 1000" or "shirts under INR 1000" returns the same price-filtered results as "shirts under 1000." The keyword "at" is also now recognized alongside "around" for price-based searches, such as "shirts at 5000." This gives shoppers more flexibility in how they express price constraints while searching, improving the accuracy of results for price-sensitive queries. |
03-Sep-26 |
|
|
Visual LLM Enrichment Configuration for the Catalog |
Multi-modal LLM enrichment can now be enabled directly through site configuration, making it easier to enhance catalog products with visual attributes. Once enabled, a set of models is pre-selected automatically, covering visual and text-based enrichment along with output validation, minimizing manual setup. Enrichment can be scoped to specific categories using the standard category picker, including root categories with all their child categories, or run across the entire catalog, with generated attributes reviewable in the Enrichment UI before being applied. |
03-Sep-26 |
Bug and Support Fixes
The following issues have been fixed in the release version 26.18
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Enterprise Dashboard: Fixed Dynamic Experience Not Triggering on Repeated Script Loads |
We have resolved an issue where loading the client tracking script multiple times on the same page could prevent dynamic experiences from triggering. In this scenario, a repeat load would reset the underlying configuration before the initial load had finished processing, causing the experience to fail silently. The tracking script now guards against repeated loads, ensuring configuration is set once and dynamic experiences trigger reliably even when the script is included more than once on a page. |
03-Sep-26 |
|
|
Enterprise Dashboard: Fixed Stale Data in Variation Options Sidebar After Save |
We have fixed an issue where reopening the "Edit Variation Options" sidebar for a variation, after saving a change, displayed the old value instead of the one just saved. The sidebar now correctly reflects the newly saved value immediately, without requiring a page refresh. |
03-Sep-26 |
|
|
Recommend: Fixed Advanced Merchandising Rule Attribution When System Filters Restrict Selected Products |
We have fixed an issue where an Advanced Merchandising rule configured to recommend specific products, with system filters and rules enabled, could return only backfill recommendations while still being attributed to the Advanced Merchandising rule. Recommendations returned in this scenario are now correctly attributed to the underlying strategy providing them, rather than to the Advanced Merchandising rule. |
03-Sep-26 |
|
|
Fixed Error When Toggling Strategy Rules from the List View |
We have fixed an issue where toggling a strategy rule on or off directly from the Strategy Rules list resulted in an "Unexpected error" and the rule's environment status did not update. Strategy rules can now be toggled reliably from the list view without needing to open the rule individually. |
03-Sep-26 |
|
|
Enterprise Dashboard: Fixed Console Error in Dynamic Experience Evaluation |
We have fixed an issue where a missing null check in the client tracking script could cause a JavaScript error ("Cannot read properties of undefined") when evaluating suitable dynamic experiences on certain pages. This has been resolved to prevent the error from occurring. |
03-Sep-26 |
|
|
Find: Fixed Region Recommendability Fallback Behavior |
We have fixed an issue where products without a recommendability mapping for a specific region could still appear as recommendable, based on the default region's recommendability rather than the requested region's. A new site configuration option, "Use default region's recommendable flag as fallback," now allows this fallback behavior to be turned off, so that products without an explicit region-product mapping are correctly treated as not recommendable in that region. |
03-Sep-26 |