Release Summary 26.22 | Oct 29, 2026

The following key features and improvements, along with bug fixes, have been released in ADA Global Digital Experience Personalization (DXP) in the release version 26.22.

Social Proof

Social Proof Performance Across All Campaigns

A new report on the Social Proof list page shows Social Proof performance across all campaigns. This report shows overall visits and metrics across all campaigns, alongside visits where Social Proof was shown and visits where it was not shown . You can filter by date range, channel, and currency, and use a toggle to switch between shown and not shown visits. The report includes daily data for trends and the number of unique products that displayed a Social Proof message.

Jira: ENG-32731

Region as a Context Parameter for Social Proof Optimization

Merchandisers can now select region as a contextual parameter for Social Proof optimization, alongside price, segment, category, channel, and day of the week. Once enabled, optimization uses the shopper’s region, passed as an API parameter or in r3_common, to choose the messages that perform best.

Messages can now be optimized by region, making them more relevant to shoppers.

Jira: ENG-33351

Discover

API Call Count for New Discover in the Usage Report

The API usage report in the dashboard now includes the new Discover method, find/v1/browse, so you can see how often New Discover is called. The report shows the total call count for the method, and you can view it across a date range using the preset options or a custom selection. A trend chart shows how call volume changes over the selected period, and the table lists find/v1/browse alongside the existing methods for easy comparison.

Merchandisers get a clear view of New Discover usage over time and can compare it with other API methods, which helps with usage tracking and planning.

Jira: ENG-33456, ENG-33335, ENG-33477

Engage

Browse and Search Contexts and Segments in Engage Campaigns

When you select a context in an Engage campaign, the full list of contexts now appears right away, so you can scroll through it and choose one. As you type, the list narrows to the contexts that match your text. Segments work the same way.

Marketers can find and reuse existing contexts and segments quickly, even when they do not remember the exact name.

Jira: ENG-33139

Ensemble AI

Accurate Bottom Lengths in Virtual Try-On

Virtual try-ons and outfits in Ensemble AI now use the catalog model image as the length reference for skirts, pants, shorts, and other bottoms. The hemline in the rendered outfit falls at the same body point as in the model image, such as above the knee, mid-calf, or ankle. When a bottom has no usable model image, the product image is used as before. The model image reference is controlled by a configurable parameter that can also be mapped to a catalog length attribute, so the model image and the attribute are used together to match length.

Shoppers see bottoms at their true length, so outfits match what they receive, which builds confidence and reduces returns.

Jira: ENG-33431

Enterprise Dashboard

Bulk Product ID Entry in Context and Segment Builders

You can now add many product IDs at once in the Context builder and the Segment builder. When you use the Product condition, a new Multiple product input option lets you paste a comma-separated or new-line-separated list of IDs, and one condition is added for each valid ID. Known products appear with their names and IDs, and duplicates are skipped. A warning shows how many IDs were not found, with a Copy ids option so you can review them. Only known products are saved. Single ID entry works as before.

Marketers can build conditions with hundreds of products in one step, for example when targeting a large set of products in a Social Proof campaign.

Jira: ENG-33357

Other Feature Enhancements

The following feature enhancements and upgrades have been made in the release version 26.22.

Jira #

Module/Title

Summary

General Availability

ENG-30761, ENG-31792

Social Proof:

No Message Option in Social Proof Optimization

Social Proof optimization can now decide that showing no message is the best choice. “No Message” is added as an option alongside your message variations and is tested and compared on the same basis as the others. If a message performs worse than showing nothing for the selected KPI, message will not shown on the product, while testing of all options continues. If a message performs better, it is shown as before. This will be applied to each variation enabled for optimization, and will be enabled through optimization configurations.

When several messages are eligible, only those that perform better than no message are shown. When “No Message” is the best option, shoppers see no message. This choice is tracked like any other message and appears in message level reporting, so you can compare it with your messages and see its effect on optimization. Merchandisers can avoid showing messages that do not help, so it will improve overall impact.

29-Oct-26

ENG-32621

Enterprise Dashboard:

Recs Response Builder

The new Recs Response Builder page lets users with the Manage Placements and Layouts role extend the recsForPlacements response by defining its structure. You can create a configuration, add keys with a name, data type, allowed values and default value, and then edit or delete keys and allowed values. A live JSON preview updates as you work. Changes stay in draft until you publish, so the live response isn’t affected while you’re editing. If a change can’t be applied, a message lists the affected placements. Before a breaking change is saved, you’ll be asked to confirm a new version.

You can now set up and maintain these configurations directly in the dashboard, without going through ADA Global.

29-Oct-26

ENG-33204

Ensemble AI: Insights on Outfits and Virtual Try-Ons

A new table in Ensemble AI shows outfit and virtual try-on activity for each style and date. For each style, it lists the number of seeds, outfits generated, seeds with virtual try-ons, and outfits with virtual try-ons, along with the site and style names.

Merchandisers can see how virtual try-ons are being used, identify high-performing styles, and decide which styles to expand or promote.

29-Oct-26

ENG-33433, ENG-33434

Ensemble AI:

More Consistent Ensembles from Day to Day

Ensemble AI now produces the same ensembles from one day to the next when nothing has changed. Products with identical scores keep the same ranking, and free form LLM styles remove duplicates in a consistent order. This applies to all sites or only to sites that use virtual try-ons, based on a setting. Ensembles are still updated when the catalog or scores change.

This reduces unnecessary new outfits and virtual try-on images, and gives shoppers a stable experience.

29-Oct-26

ENG-33339

Autocomplete for Layout Selection

When you add a layout to a placement, the layout selection window now suggests matching layouts as you type. This works on the placement list page and on the new placement profile page.

Finding the right layout is faster, especially for sites with many layouts, such as email layouts.

29-Oct-26

ENG-33478

Enterprise Dashboard:

Product Analytics Now Sent to Mixpanel’s EU Servers

The web app and documentation now send product analytics directly to Mixpanel’s EU servers, matching where the data is stored. Existing events, such as page views and clicks, keep the same names and details.

This keeps usage data flowing into dashboards without interruption.

29-Oct-26

PLAT-4470

Platform:

Same Enrichment Attribute Names Across Sites

Different sites can now define enrichment attributes with the same name, such as “color”, without conflict, and each site’s attributes are treated independently. Attribute names remain unique within a single site. Each calculation is associated with one site, and a site is required for property schema configuration. Validation for other vendors is unchanged.

This lets each site set up its enrichment attributes freely, without renaming to avoid clashes with other sites.

29-Oct-26

PLAT-4474

Platform:

More Reliable Enrichment Processing Under High Load

The enrichment engine now keeps processing steadily when a large backlog of items builds up across multiple partitions, and it reports its health accurately. Newly ingested items are added to the enrichment index as they arrive.

Enrichment stays dependable during heavy ingestion, so product data is updated without delays.

29-Oct-26

PLAT-4475

Platform:

No Item Loss During Service Restarts

Catalog updates are no longer dropped when the service restarts during deployment or maintenance. If the property definitions cannot be loaded because of a restart, the update is held and processed again instead of being dropped.

Your catalog stays complete and accurate through deployments.

29-Oct-26

PLAT-4479

Platform:

Reliable Product Lookups with Missing Numeric Values

Product lookups now handle products that have an empty value for a numeric attribute, such as sale price. These products are processed with default values instead of causing an error, so they can be returned in recommendations.

Recommendations stay complete and reliable even when some catalog data is incomplete.

29-Oct-26

PLAT-4485

Platform:

Faster and More Accurate Catalog Ingestion

The streaming engine now handles catalog updates more efficiently, so large batches are processed faster without delays. Item counts, including recommendable and non-recommendable products, stay accurate when batches are processed in parallel, and they are updated correctly when items are added or deleted.

Catalog updates are applied sooner, and item counts remain reliable.

29-Oct-26

PLAT-4471

Platform:

Faster Recommendation Processing with Lower Memory Use

The recommendation server now uses less memory and processing time when evaluating categories, products, placements, sorting, and localized attributes. Recommendations behave exactly as before, including for sites that use regions.

Responses are faster and the system handles higher traffic more efficiently.

29-Oct-26

PLAT-4480

Platform:

Faster Region-Based Product Lookups

Region-based product lookups are now faster, which improves recommendation response times for sites that use regions. Region filtering, pricing, and availability work as before, and the update is compatible with existing catalog models.

Shoppers get faster, more responsive recommendations on region-based sites.

29-Oct-26

Bug and Support Fixes

The following issues have been fixed in the release version 26.22

Jira #

Module/Title

Summary

General Availability

ENG-33389

Enterprise Dashboard:

Localized Text Preserved When Duplicating Guided Selling Experiences

We have fixed an issue where duplicating a Guided Selling experience with localized text showed variable names instead of the original text and left the Translation tab empty. Duplicated experiences now keep all text in the start and quiz screens, along with their translations.

29-Oct-26

ENG-33501

Recommend:

Scheduled DSW Jobs with the Same Name Now Run

We have fixed an issue where scheduled DSW jobs that shared the same name did not run. Each job now runs at its scheduled time and is tracked separately in its own status history.

29-Oct-26