Google’s Universal Cart reached US shoppers on October 15, and the four trade accounts that followed it describe four different products: a data question, a retailer roster, a piece of plumbing and a retailer standoff. Read side by side, they leave one fact for marketers unstated, which is when a retailer first learns that a shopper saved its product.
What Google launched
Suresh Ganapathy, Senior Director of Product Management for consumer shopping at Google, introduced the cart in a company blog post on October 15. He summed it up in one sentence: “The cart puts you in control, helping to organize your shopping, keep track of items you love from across retailers, stay on top of promos, and enjoy effortless checkout powered by the Universal Commerce Protocol (UCP).”
The cart is live in the US across Google Search and the Gemini app. It runs on the Shopping Graph, which Google says holds “60+ billion continually refreshed product listings”, and it gives shoppers two ways to finish: check out with Google Pay inside the cart, or transfer the items to the merchant’s own cart. Google says that on either path shoppers get confirmations and support from the merchant, who will “remain the seller of record.” Nike, Sephora, Target, Ulta Beauty, Walmart, Wayfair and many Shopify merchants are named as offering the simplified checkout experiences.
Merchants do not simply switch it on. Google’s UCP developer guide says “Your integration must be approved by Google before you can go live on AI Mode in Google Search and Gemini,” which makes the roster a list of approved integrations. That matters when reading the coverage below.
Four accounts, four emphases
MediaPost took the advertising angle. Laurie Sullivan wrote on October 15 that the cart “changes ad targeting and where the data sits.” Her argument is that products left in a Universal Cart sit in Google’s ecosystem, which shifts control of retargeting and cart-abandonment data away from the retailer’s own site. She also wrote that it is unclear whether Google or the retailer is responsible for triggering a retargeting ad for items left behind, and she noted that Google’s UCP guide keeps merchants as the merchant of record. Debra Aho Williamson, founder and chief analyst of Sonata Insights, is quoted from a newsletter published that day: “Both approaches will reveal a lot about the future role of agents in shopping, the types of purchases consumers are and aren’t willing to delegate, and how seamless (or not) the handoff will be between consumers, retailers and payments systems.”
Digital Commerce 360 took the retail angle. Brian Warmoth’s October 16 report lists the launch retailers, notes that Newegg added Google UCP checkout earlier in October, and walks through the features Ganapathy described. It frames the cart as a shopper tool timed to the holiday season and says nothing about ad targeting or retargeting data.
eWeek took the engineering angle. Liz Ticong’s October 16 piece reviews Google’s documentation and concludes that saving products together and keeping retailer carts synchronized are separate capabilities. Native checkout can buy multiple products from one participating retailer in a single transaction, but adjusting quantities inside Google’s interfaces is still forthcoming. The separate Cart API supports a one-way transfer to the retailer’s site, and its first release does not synchronize later changes between Google and the merchant’s cart.
Startup Fortune took the retailer-resistance angle. Ron Patel wrote on October 17 that Walmart, Target and Sephora support only select checkout options rather than a full handoff, and read that as retailers protecting their customer data while still wanting the traffic. That reading is the outlet’s own. Google’s post lists those retailers among the merchants offering simplified checkout and does not say which options each one supports.
Where the accounts disagree
The sharpest disagreement is about what is live. Startup Fortune describes a shopper adding a graphics card to the same cart while watching a YouTube review. Google’s post names Search and the Gemini app only, and eWeek reports that an earlier I/O preview promised Gmail and YouTube support, with no date given in the October 15 announcement. A reader of the Startup Fortune version would expect a surface that Google’s own launch post does not include.
The second disagreement is about how open the retailers are. Google’s post and Digital Commerce 360 present a broad, willing roster. Startup Fortune presents a guarded one. The primary documents cannot settle it, because Google’s post names merchants without naming the checkout option each supports. The accounts also differ in what they treat as the news. MediaPost leads with data location, eWeek leads with synchronization limits, Digital Commerce 360 leads with features, and Startup Fortune leads with who gets squeezed.
They agree on three things: the cart launched October 15, it covers Search and Gemini, and merchants stay the seller of record.
What none of them says
Put the MediaPost and eWeek accounts together and a gap appears. MediaPost says saved-item data sits in Google’s ecosystem. eWeek says the Cart API is a one-way transfer that does not synchronize later changes. None of the four accounts, and nothing in Google’s launch post, says whether a retailer is told when a shopper saves its product, or only when the shopper chooses to transfer or buy.
In our reading, that gap decides how much of the shopper’s behavior a retailer can see. A save, a wait for a price drop and a return visit are the events that abandonment emails and retargeting audiences are built on. If those events happen inside Google’s cart, a retailer’s own tags may record nothing until the transfer click, and a shopper who buys through native checkout may arrive in the retailer’s systems as a finished order with no browsing history attached.
The same logic applies to measurement. We covered how agentic shopping surveys disagree because they ask different questions, and the cart coverage repeats the pattern: each outlet answered the question closest to its own readers. Our earlier report on agent checkout in testing at TikTok and the consolidation of Amazon’s ad console and DSP into one agent show the same direction from other platforms: the place where a shopper decides is moving onto surfaces the retailer does not run.
What it means for the marketing leader
Start with an inventory of the cart events your programs depend on. List every email, audience and report that keys on an add-to-cart, an abandoned cart or a returning cart session, and mark which ones are fed by tags on your own site. Those are the ones a Universal Cart save will not trigger by default.
Ask your Google contact and your commerce platform vendor to state what your reporting shows when a shopper saves one of your products to Universal Cart, and what it shows when they transfer or buy. Google’s Merchant Center and your analytics tool may answer differently, and none of the four accounts gives the answer.
Decide which checkout path you want. Native checkout keeps the shopper inside Google’s surfaces. Transfer sends them to your site with the items loaded. The eWeek piece relays a Walmart executive’s earlier remark that purchases completed inside ChatGPT converted at about one third the rate of products that sent shoppers to Walmart’s site, and the same article cautions that differences in product selection stop that gap being pinned on checkout design alone. The figure is one retailer’s experience on a different platform, so treat it as a reason to test, not as a forecast.
Finally, capture identity at the first moment you can. A transferred cart lands on your site with a shopper you may not recognize. A login prompt, a loyalty prompt or an email capture at that step is where first-party data can be rebuilt.
Before Black Friday, buy one of your own products through Universal Cart on both paths and write down what your analytics, your ad platform and your order system each record at every step. That list is the only reliable answer to the question the four accounts leave open.
Source: Google Shopping blog
