Social Rocket
Google continues to expand its shopping and commerce infrastructure for AI search. New features include AI Performance Insights in Merchant Center, a Business Agent in YouTube Ads and additional functions for the Universal Commerce Protocol (UCP) checkout. The most important updates were presented on 16 September 2026 as part of Google's Rethink update.
For retailers, the new visibility is especially interesting: Merchant Center is meant to show how brands and products are discovered in AI Mode and AI Overviews. However, the feature is not yet available in Germany at launch.
AI Performance Insights measure visibility in AI Mode
Google is making AI Performance Insights in Merchant Center generally available for businesses in Australia, Canada, India, New Zealand and the US.
According to Google, the tool shows, among other things, how strongly a brand or its products appear on AI surfaces compared with competitors. AI Mode and AI Overviews are explicitly named.
With this, Google is trying to close a gap that has been hard for many marketers to measure: products can appear in generative search experiences, yet classic search reports do not explain how often this happens or in which competitive context.
For German retailers, the feature remains for now a preview of what is likely to become more relevant once Google expands coverage. Google currently does not name a German launch.
Business Agent comes to YouTube Ads
Google is also bringing its Business Agent into a beta for YouTube Ads in the US.
The agent is embedded directly in ads. Users should be able to ask complex product questions and receive individual answers without leaving the YouTube context.
This changes the role of the ad: it is no longer just a click path to a shop, but can itself become an interactive advisory surface.
The approach fits a broader trend in which social, video, search and AI are moving closer together in the purchase process. Srocket recently classified this trend in Shopping between social search and AI and in the new IAB study on creator voices in AI shopping recommendations.
UCP checkout gains more functions
In parallel, Google is expanding the Universal Commerce Protocol (UCP) hub in Merchant Center.
UCP is intended to allow retailers to process purchases directly through Google surfaces. According to Google, new additions include, among others:
- cart handover to the retailer's shop,
- expanded checkout tests,
- further analytics functions in preparation.
The new functions are being rolled out gradually in the US. Australia and Canada are set to follow in early 2027.
Google names Tapestry, the company behind Coach and Kate Spade, as an example of a retailer using UCP-based checkout in Search, AI Mode and the Gemini app.
Product feeds become more important for AI search
The classic Merchant Center feed remains central. Google continues to recommend clean product data to retailers, but increasingly supplements it with information that is useful for conversational systems.
This includes, for example:
- video link attributes,
- additional conversational product attributes,
- loyalty and membership data.
In a test with lululemon, according to Google, submitted conversational attributes were used in 50% of relevant product recommendations in AI Mode. This is a company test and not an independent market study, but it clearly shows the direction Google is pursuing: structured product information should feed not only Shopping Ads but also generative product recommendations.
Google also reports that retailers who implement central Merchant Center feed recommendations see on average 5% more conversions in the following month. This figure also comes from Google's own data and should be classified accordingly.
What this means for German retailers
The most important development is less a single feature than the architecture behind it. Google is increasingly connecting:
- product data,
- AI visibility,
- advice,
- ads,
- checkout.
This makes the Merchant Center feed an even more important data source for discovery and conversion. Those who maintain product information only for classic Shopping Ads may be thinking too narrowly.
For German marketers, the immediate takeaway is mainly this: feed quality, structured product information and clear product benefits should already be maintained in a way that can also work on conversational surfaces. The new AI Performance Insights themselves are not yet available in this country.
Other platforms are also moving commerce surfaces closer to discovery and recommendation. Pinterest, for example, is opening its visual search more strongly to advertising with Visual Search Ads, while Walmart is expanding its retail media offering more broadly across CTV, social and offsite with Walmart Connect.