Google launches Universal Cart: What this means for brand and product visibility

Google made Universal Cart available to US users on 15 October, across Search and the Gemini app. Shoppers can now gather products from different retailers into one cart, get alerts when prices drop or items come back in stock, and check out through Google’s Universal Commerce Protocol (UCP).

Because the cart runs on the more than 60 billion listings in Google’s Shopping Graph, every product Gemini recommends can now end in a sale. That makes Agentic Commerce Optimisation (ACO), the practice of measuring and improving how LLMs and AI shopping agents find, interpret and recommend products, a revenue priority for brands.

It also raises a harder question about where brands should focus. Azoma’s analysis of citation data across our commerce client dashboards shows that Gemini and AI Overviews, Google’s two main AI shopping surfaces, draw on near-opposite sources, so a brand can win on one and be invisible on the other.

What is Universal Cart?

Universal Cart is a single shopping cart that follows the shopper across Google. A product saved from a Gemini conversation, an AI Mode answer or an ordinary Search result lands in the same cart, even when the products come from different stores. Google first showed the feature at its I/O developer conference in May 2026, and the US launch makes it available to everyone there.

Once items are saved, Google keeps working on the shopper’s behalf, flagging price drops and restocks and surfacing promo codes and card perks through Google Wallet. When the shopper is ready, they can pay with Google Pay inside the cart or move the items to the retailer’s own site. Nike, Sephora, Target, Ulta Beauty, Walmart and Wayfair are among the merchants taking part, and in every case the merchant stays the seller of record and handles customer support.

Where does UCP fit in?

The checkout runs on UCP, the open standard Google introduced earlier this year so that AI agents and retailers can work together to checkout. It is what lets Gemini take a shopper from a recommendation to a completed order in one flow.

That turns product data into a checkout dependency. When an agent sends a shopper to pay, it passes the product’s variant ID as the line item, so a missing or mismatched variant ID, SKU or GTIN stops the sale, as our Universal Cart and Target guide explains. Our post on Gap’s checkout inside Gemini looks at an early retailer integration.

How to enable UCP checkout on Google

Google runs onboarding through a UCP integration hub in Merchant Center, which is rolling out in the US now, with Australia and Canada to follow early next year. Brands can offer transfers to merchant carts, where shoppers add products on Google and finish on the brand’s own site, or native checkout on Google, where they pay without leaving Google.

1. Prepare Merchant Center: Complete your shipping, returns and product feed settings, and make sure the product IDs in your feed match those in your checkout system, as Google’s UCP guide sets out.

2. Open the hub: In Merchant Center, go to Settings, then UCP integration.

3. Build and publish your UCP profile: Host it at /.well-known/ucp. Then implement the Cart API for cart transfer, or, for native checkout, three REST endpoints that create, update and complete checkout sessions, with Google Pay handling payment.

4. Test in the sandbox:  Share separate API keys for staging and production, then simulate cart and checkout actions and review your server’s responses.

5. Validate in production: Select capabilities that match your profile and complete the checklist, including order-status webhooks for native checkout. Then agree to the UCP Program Addendum, and Google’s support team reviews the integration before it goes live.

Cart transfer is the faster route, since it skips full product feed setup, product warnings and customer support details. That makes it a sensible first step for most brands.

Why does this matter for brand and e-commerce teams?

Universal Cart folds discovery, comparison and checkout into the AI answer, so a shopper can go from question to order without ever opening a product page. The recommendation has become the shelf.

The audience is already there. Google says AI Overviews has more than 2.5 billion monthly users and AI Mode has passed 1 billion, and with a checkout now behind those answers, a brand that Gemini leaves out loses the sale as well as the mention. Price-drop alerts add pressure, exposing an uncompetitive offer the moment a shopper saves it.

This is the shift the industry’s vocabulary is catching up with. Generative Engine Optimisation (GEO) and Answer Engine Optimisation (AEO) describe getting into the answer, and ACO covers what Universal Cart now makes possible on Google, where the answer ends in a purchase.

How can brands get recommended in Gemini and Google AI Overviews?

Tracking shows where you are missing, and the 5Cs of Agentic Commerce explain why. Developed by Azoma with the Digital Shelf Institute from Azoma’s citation data, they describe the five things an AI shopping agent checks before it recommends a product. On Google, each one maps to something specific: your Merchant Center feed, your product pages, the sources Gemini cites, or your UCP setup.

Completeness: Is the product data there?

Completeness means auditing product data the way each agent actually reads it. AI agents judge products on backend metadata rather than marketing copy. In Azoma’s data, a single listing on a global retailer can need more than 600 attribute values before an agent treats it as complete. On Google, that data lives in Merchant Center.

Start by filling core attributes on every active SKU, including mid-tail and long-tail lines. Then add the structural fields among Google’s conversational attributes:

• item_group_title and variant_option describe variant families.

• related_product links accessories, required parts and substitutes.

• document_link adds manuals and spec sheets.

• popularity_rank flags your best sellers within your catalogue.

• question_and_answer provides user-, merchant-, and manufacturer-authored questions and answers about the product. 

All are optional and complement your existing feed.

Context: Does the product answer the shopper’s question?

Context means answering the questions shoppers actually ask AI. A complete feed can still lose if nothing says who the product is for, where it is used or what it cannot do. Audit product pages against the long-tail prompts in your category and spell that detail out, such as “fits a 15-inch laptop” or “suitable for sensitive skin”.

Google’s AI performance insights show the product terms shoppers use, such as “easy setup” for tents or “long battery life” for portable lamps. That tells you which language to add. In Merchant Center, the question_and_answer attribute carries this context straight into AI answers, with up to 30 question-and-answer pairs per product. Build them from real shopper questions, support tickets and reviews, and leave out anything already in your description, product highlights or product details.

Citations: Do the sources each surface trusts mention you?

Citations means showing up on the right off-platform surfaces, and Google’s two AI surfaces trust very different ones. In Azoma’s September 2026 citation analysis, earned media such as reviews and editorial roundups made up 51.9% of Gemini’s citations but only 7.6% of AI Overviews’. Retailer pages ran the other way, at 15.5% of Gemini’s citations and 39.2% of AI Overviews’.

Share of citations by source type, September 2026

Source: Azoma platform data across commerce client dashboards.

Put simply, Gemini behaves like a shopper reading reviews and gift guides, while AI Overviews behaves like one browsing retailer listings. A product with strong retail listings can appear in AI Overviews and still be missing from Gemini if the publications Gemini trusts have not covered it. These shares describe citations across many brands, so each brand’s own mix will differ.

The work splits accordingly. Earn coverage in the reviews, gift guides and roundups Gemini cites, strengthen retailer listings for AI Overviews, and keep product facts identical across your site, Merchant Center and retailer feeds.

Correctness: Is what the agent says true?

Correctness means catching what AI gets wrong and correcting it at scale. AI agents generate answers by prediction, so they can state wrong specs even when a brand’s own content is 100% accurate. Monitor Gemini and AI Overviews for wrong specs, prices or claims, trace each error to the page the answer cited, and correct it at the source.

Customer Acquisition: Does visibility turn into customers?

Customer Acquisition means connecting AI readiness to revenue, and on Google that happens through UCP. Merchants must support guest checkout, which Google describes as having fewer steps and being optimised for conversion. As it rolls out, identity linking adds account sign-in through OAuth 2.0, so a shopper’s loyalty benefits, personalised offers and signed-in checkout carry into Gemini and AI Mode, as Target did with Target Circle.

To offer identity linking, declare it in your UCP profile and publish your OAuth endpoints. Google’s optional Streamlined Linking keeps the whole sign-in flow inside Google’s interface. Once live, the UCP Analytics dashboard in Merchant Center follows the funnel from first click to completed purchase.

How can brands measure their visibility on Gemini and AI Overviews before the holidays?

Measurement comes first, and the window before peak trading is short. Start this week with your top 20 holiday SKUs, write 5 to 15 prompts for each in the language shoppers use, and run them across Gemini, AI Mode and AI Overviews, noting which products are named, how they are described and which pages are cited. Re-run the same prompts weekly through December, since source mixes move quickly and Universal Cart’s price alerts make stock and price part of the picture.

Doing this across a full catalogue, SKU by SKU and surface by surface, is where a dedicated platform helps. Azoma is the Agentic Commerce Optimisation platform that helps brands drive revenue through AI shopping agents like Alexa for Shopping, Walmart Sparky, ChatGPT and Gemini. It also tracks Google AI Overviews, and works with global consumer brands including L’Oréal, Unilever, Mars, Beiersdorf and Reckitt.

What Universal Cart does not change

Universal Cart adds a checkout and leaves the recommendation logic where it was. Gemini still assembles its answers from the sources it cites, and Google’s launch post describes no new way for brands to place products in the cart. Tracking has limits too: prompts sample answers and cannot see each shopper’s personalised conversation, so visibility gains are best confirmed against conversion data.

Summing up

Universal Cart is Google’s clearest signal yet that AI answers are becoming storefronts. With checkout built into Gemini through UCP, the recommendation is now the sale, and Gemini and AI Overviews each decide on a different mix of sources.

Google is not alone. OpenAI’s Agentic Commerce Protocol, retailer assistants such as Alexa for Shopping and Walmart Sparky, and a new wave of personal AI agents are all moving the purchase into the answer, each drawing on its own sources. Brands that treat each agent as a separate silo will spend more and see less, which is why Azoma measures and optimises across all of them in one place.

The brands that do well will know, SKU by SKU, what each AI assistant says about them and why, and will fix the sources behind every miss. 

Read our 5Cs whitepaper for the full approach, or book a call with the Azoma team to see where your products stand today.

FAQs

What is a Gemini brand visibility tracker? A Gemini brand visibility tracker runs shopping prompts through Google’s AI surfaces on a schedule and records how a brand and its products appear in the answers. Useful trackers log the products named, their position, how each is described, which competitors appear and which sources are cited. Tools such as Azoma report the Gemini app and AI Overviews separately, because the two cite different sources.

How can I monitor what Gemini says about my products? Write 5 to 15 shopper-style prompts per priority SKU, run them on a fixed schedule under recorded conditions, and log every answer product by product. Compare the results with Merchant Center’s AI performance insights and Search Console’s generative AI report. For a large catalogue, a tracking tool such as Azoma automates the runs and keeps the history comparable over time.

What tools help brands get recommended in Gemini shopping results? Monitoring tools show where a product is missing and which sources Gemini cited instead. Getting recommended then depends on fixing those sources: complete retailer and Merchant Center attributes, coverage in the editorial roundups Gemini cites, and brand pages that answer real shopper questions. Earned media accounted for 51.9% of Gemini citations in September 2026.

Is there a platform that tracks AI Overviews and Gemini recommendations together? Yes. As of October 2026, Google’s own Merchant Center AI performance insights report share of voice across both, and several AI visibility trackers and SEO-suite add-ons cover them too, as does Azoma. Check that results for the two surfaces are reported separately, because in Azoma’s data Gemini cited earned media nearly seven times as often as AI Overviews did.

Do brands need UCP to be recommended in Gemini? Being recommended and being buyable inside the answer are separate steps. Gemini’s recommendations are assembled from the sources it cites, while UCP lets an eligible product be bought inside Gemini or AI Mode, with the merchant remaining the seller of record. Brands that sell direct can work on both; brands that sell through retailers depend on those retailers’ UCP integrations.