Google's Universal Cart keeps the save. Stop treating add-to-cart as your signal

Every ecommerce team we know has a dashboard tile for add-to-cart rate. It's one of the first numbers people check on a Monday, and it feeds the email flow that nudges a shopper who left a moisturizer or a case of sparkling water sitting in the basket. Google just moved that moment somewhere a store can't see it.

Universal Cart launched in the US on October 15 across Google Search and the Gemini app. A shopper can save a product from one store in AI Mode, another from a second store in Gemini, and find both waiting later. Google's post says the cart is built on the more than 60 billion listings in its Shopping Graph, sends alerts when prices drop or items come back in stock, suggests which card to pay with because it sits on Google Wallet, and surfaces promo codes. When the shopper is ready, they either pay with Google Pay inside the cart or transfer the items to the store's own cart. Search Engine Roundtable confirmed on October 23 that it's showing up live in Search.

Most of the coverage has been about shoppers. The more interesting question is what the store sees.

What reaches the store, and when

The best account we've read comes from Brandon Lazovic, an ecommerce SEO consultant who went through Google's launch post and eight developer guides, and whose findings PPC Land laid out on October 16. The short version: nothing reaches the store until the end.

Google's Cart API reference, as Lazovic quotes it, says Google "accumulates those items locally" and doesn't fire calls as items are added. The merchant's cart endpoint gets one request with everything in it, only when the shopper clicks transfer. The current release supports creating and transferring a cart, and nothing else: no updates, no syncing, no deletes. Merchant Center's reporting for this traffic begins at a metric called "Clicks to Buy."

Shopper does	Store receives	Shows in Merchant Center reports?
Saves your product in AI Mode or Gemini	Nothing	No
Gets a price-drop or restock alert	Nothing (Google uses its copy of your data)	No
Edits or removes items	Nothing (no sync in this release)	No
Presses transfer	One cart request with every saved item	Yes, from "Clicks to Buy"
Checks out with Google Pay on Google	Order, via the native checkout endpoints	Yes

Lazovic is careful to say he read this from documentation and didn't run a live transfer himself, so treat the table as the documented design, not a measured result. But the design is clear enough. As he puts it, a product saved on a Monday and transferred on a Friday arrives as "a Friday visitor who arrived with a full cart." A product saved and never transferred "appears in no report you can open."

The obvious reaction, and why we don't share it

The first reaction in most marketing teams will be loss. The save is an intent signal. It powers remarketing lists and the abandoned-cart sequence, which for a lot of DTC brands is one of the few emails that reliably earns its keep. Now Google holds it, and the merchant sees only the bottom of the funnel. Google holding the most complete picture of which products shoppers are weighing across stores is also exactly the data that would make cart-stage advertising valuable, a point PPC Land makes and we think is fair.

We'd push back on the mourning, though, for two reasons.

The first is that add-to-cart was never as good a signal as the dashboards suggested. People use carts as wish lists, price trackers and bookmarks. Anyone who has run an abandoned-cart flow knows how much of the "recovered" revenue would have come back anyway. A cart that now does the price tracking for the shopper, across stores, takes away a behavior that was noisy to begin with.

The second matters more for CPG. For most food, beverage, beauty and household brands, the cart was already somewhere else. A shopper buying your granola most often puts it in a Walmart, Target, Amazon or Instacart basket, not on your Shopify store. You never saw those saves either. Universal Cart moves the "maybe" moment from one place you couldn't see to another place you can't see. For the share of sales that runs through your own site, it's a real change. For the brand as a whole, it's less dramatic than the headlines.

The save moved from one place a CPG brand couldn't see to another place it can't see. What moved into view is how much Google's copy of your data now matters.

What actually changed: Google's copy of your data is now acting on shoppers

Here's the part we'd spend time on. Universal Cart isn't passive storage. It watches prices and stock and tells shoppers when something changes, and it does that from Google's copy of your product data, not from your live site.

That puts a lot of weight on feed accuracy. Lazovic flags one risk as "suspected, not confirmed": a sale price that has ended on your store but still sits in Google's copy could trigger a price-drop alert, sending a shopper to a higher price than the notification promised. Nobody has shown that happening yet. But anyone who has watched a Merchant Center feed lag a promotion by a day or two can picture it, and for, say, a $6 item a shopper was told costs $4.50, the disappointment lands on your brand, not on Google.

Promotions work the same way. Lazovic reports that Google's checkout guide describes "auto-applied promotions" as offers taken directly from Merchant Center feeds, and allows shoppers to type in up to 10 codes per session. Where the in-cart promo codes come from beyond that isn't documented. If your holiday offer only exists as a site banner, the cart may not know about it. If it lives in Merchant Center, it can travel with the product. We made the same point about AI Mode shopping ads in our holiday playbook, and it applies twice over here.

Native checkout is opt-in, unless something opted you in

The other decision is whether shoppers can buy your products without leaving Google. Google's Merchant Center help page on UCP says only listings using the native_commerce(checkout_eligibility) attribute show the Buy button for this checkout, and the merchant stays the seller of record. Lazovic reports that the attribute defaults to false, and that subscriptions, personalized goods, final-sale and pre-order items and age-restricted products are excluded from native checkout altogether.

That sounds like you'd have to choose in. Shopify stores should look more closely. Shopify's agentic storefronts settings say that by default every available channel gets your products, including Google AI Mode and Gemini through the Google & YouTube channel or Merchant Center, and that direct checkout is active where supported. If nobody on your team has opened Sales channels, then Agentic, the default made the choice for you. That isn't necessarily wrong. A consumable that people rebuy is exactly the product where a two-tap checkout makes sense. It just should be a decision rather than an accident, and we've written up what to check before Black Friday.

Which launch merchants were visibly ready

One more detail from Lazovic's work is worth knowing, with heavy caveats. A store that supports this protocol is supposed to publish a small public file at /.well-known/ucp listing what it can do. On October 16 he fetched that file for eight merchants Google had named.

Merchant	Result at /.well-known/ucp
Steve Madden (Shopify)	Profile found, 11 capabilities
Fenty Beauty (Shopify)	Profile found, 11 capabilities
Nike	404
Target	404
Ulta Beauty	404
Walmart	404
Wayfair	404
Quince	404

A 404 here means only that no file sat at that address for one request from one connection. Big retailers may register their profile somewhere else with Merchant Center, or block requests like his. The two Shopify stores publishing identical lists suggests Shopify generates the file for its merchants. For a CPG brand on Shopify, it's a two-minute check: open yourstore.com/.well-known/ucp in a browser and see whether it exists.

What we'd tell a brand this week

We'd skip the debate over whether Google should share saves. That may change, and it's out of your hands. What's in your hands:

• Make Google's copy of you right. Prices, sale end dates, stock and pack sizes in Merchant Center have to match your store, because they now generate alerts shoppers act on. If your feed updates daily but your promotions change hourly over Cyber Week, close that gap. Our Merchant Center tip covers the basics.
• Tag the transfer. Your server writes the continue_url Google sends shoppers to after a transfer, so you control its parameters. Lazovic recommends adding one that marks the visit as a Universal Cart transfer, because an untagged one can't be told apart afterward. Do it before the peak, not after. Our guide to finding AI traffic in analytics has the rest of that setup.
• Decide on native checkout per product. Replenishment SKUs and sampler packs are good candidates. Anything that needs a quiz, a shade match or an age check is not, and some of those are excluded anyway. See tip 33 for how we think about letting agents buy.
• Check what the assistants say before they say it to shoppers. The cart only helps if your product got shortlisted in the AI Mode or Gemini answer first. We run shopper-style questions through those assistants every month. Tools such as XENA Intelligence's SearchPanel do that across assistants and score each SKU on how it shows up, but a spreadsheet and a free afternoon will get you started.

The part we can't answer yet

We don't know whether Google will start selling placements at the cart stage, though the data would support it. We don't know when YouTube and Gmail, which Google mentioned in May, will join. And we don't know how many shoppers will use a cart that lives on Google rather than on a store. Holiday traffic will tell us more than any launch post.

What we're fairly sure of is the direction. The interesting moments in a purchase keep moving into Google's and the assistants' own surfaces, and the store becomes the place where the order gets fulfilled. Brands that treat their product data as the storefront will mind that less than brands still watching the add-to-cart tile.