Google’s cart now follows shoppers across Search, AI Mode and the Gemini app — and in some flows the payment finishes inside Google, though the merchant stays the seller of record.
What this means for you
- Universal Cart went live in the U.S. on October 15, 2026; a sale can finish inside Google, without reaching your own site.
- The handoff to your site runs one way — you cannot send the shopper back into Google’s cart.
- Owning checkout keeps the email, the data, and the upsell; a hosted cart keeps none of that for you.
- Baymard Institute puts average documented cart abandonment near 70 percent, before any handoff friction.
- A slow landing page is one way a handoff quietly fails — Baymard’s top documented reason is unexpected extra costs.
How the cart moved into Search and Gemini
Google is rolling out Universal Cart in the U.S. across Search, AI Mode and the Gemini app, per Search Engine Journal’s reporting. A shopper can drop a jacket from one retailer and a lamp from another into the same basket before paying for either.
The cart itself, not any single store, becomes the shopping session.
From that shared basket, one of two things happens next. Checkout can finish inside the cart with Google Pay, where the merchant remains seller of record and still sends confirmations and support. Or the shopper is handed off, one way, to the merchant’s own site to finish paying there. Google does not promise both routes at every store.
That’s a shift from the typical results page. Instead of a list of links to different retailers, a shopper works from one running basket Google tracks as they search, compare, and add more.
For a small retailer, that handoff is the moment that decides whether a sale actually lands. It is also the exact gap we rebuild most often at ProXSites — a storefront with traffic but no real path to a finished checkout.
Plenty of local businesses already have a Google Business Profile pulling in views. Universal Cart adds a payment layer on top of that discovery — a cart that can route a paying customer to a site that isn’t ready for them.
That’s the part worth sitting with. The rollout itself is plumbing; what it changes is how often a shopper reaches a site at all.
Checkout inside the cart, or handed off to you
The two paths work very differently for a merchant. Hosted checkout finishes the transaction inside Google’s surface. A one-way handoff sends the shopper to the merchant’s own site instead, where that page becomes the entire rest of the sale.
One-way is the operative word in that second path. The first release only creates a cart on the merchant’s site; Google does not support updating, syncing or returning a shopper to its own cart afterwards. Whatever happens next happens on infrastructure the merchant controls.
It also changes what a merchant can measure. Google says the merchant stays seller of record on both routes, but a sale completed inside Google is recorded through Google’s checkout first. A sale finishing after a handoff lands in the merchant’s own platform, where normal analytics and remarketing tools still work.
Google has tried the hosted model before. Under Buy on Google, shoppers could complete a purchase without leaving Google; Google discontinued that program in September 2023 and pushed merchants back toward their own checkouts.
That history is worth remembering. A platform’s enthusiasm for hosting checkout has shifted before, and a merchant’s own site is the one piece of the funnel that doesn’t depend on which way Google leans next.
None of that makes hosted checkout irrelevant for every business. It may suit a retailer chasing pure volume on a single low-margin product. It suits far fewer of the service businesses and local retailers running on tighter margins and repeat customers.
Why the destination still has to be fast
Speed decides a large share of what happens after a handoff. Google’s 2017 mobile benchmarks report, drawing on 2016 data it has not since refreshed, found 53 percent of mobile site visits are abandoned when a page takes longer than three seconds to load.
A handoff that lands on a slow page is a handoff that quietly fails. The shopper doesn’t complain — they just leave before the page ever finishes loading.
Core Web Vitals still set the bar Google uses to judge that landing page: Largest Contentful Paint under 2.5 seconds, Interaction to Next Paint under 200 milliseconds, and Cumulative Layout Shift under 0.1, measured at the 75th percentile of real users.
Those thresholds have not changed since INP replaced First Input Delay in March 2024, and they apply to the exact page a handoff drops a shopper onto, not just the homepage.
None of these thresholds are optional once real purchase traffic depends on them. A page that passes under normal traffic can still fail under a sudden spike, the moment a handoff sends a wave of visitors at once.
Cart abandonment was a problem long before any of this. Baymard Institute’s meta-analysis of roughly 50 studies puts the average documented online cart abandonment rate at about 70 percent.
“Once Google hands a shopper to a merchant’s checkout, there is no cart to fall back on — that one page is the entire rest of the sale.”
What ownership of checkout actually buys you
A checkout a business owns captures more than a payment. It’s the one place to ask for an email address, surface a loyalty offer, or set a retargeting pixel without asking anyone else’s permission.
Marketplaces that host checkout on their own surface have historically limited how much customer data flows back to the merchant. Data follows whoever processes payment.
A one-way handoff flips that arrangement. The merchant processes the payment, so the merchant keeps the full order record, the contact details, and the picture of what that customer buys next.
That’s exactly the layer our Pro and VIP builds are set up to capture — a working blog and SEO foundation on one tier, analytics and A/B tested pages on the other, both aimed at turning a first visit into a return one.
An email captured at checkout is worth more over a year than almost any single sale, because it’s the one channel a merchant can use again without paying a platform to reach the same customer twice.
The same logic applies even to businesses that never think of themselves as e-commerce. A clinic collecting a deposit, a contractor invoicing a job, a restaurant running its own online order form — all of it depends on a checkout page that works the moment traffic lands on it.
None of that requires a shopping cart in the traditional sense. A deposit form, a booking page, or an invoice link still functions as a checkout, and still needs the same speed and trust a handoff assumes.
What to check before the first handoff lands
Before Universal Cart sends a single paying shopper toward a merchant’s own site, there are concrete things worth checking first.
- Confirm checkout works cleanly on a phone screen.
- Check Largest Contentful Paint and Interaction to Next Paint against Google’s published thresholds.
- Capture an email address during payment, not only on a post-purchase receipt page.
- Keep the domain, SSL, and basic trust signals visible on the exact handoff landing page.
- Audit legal pages — shipping, returns, privacy — since cold traffic looks for them fast.
- Make sure the handoff page doesn’t require creating an account before it will accept a sale.
None of that happens automatically just because the traffic shows up. A handoff only pays off if the page underneath it is actually built to finish the sale.
That’s the groundwork a Starter site handles at a basic level, and the groundwork a Pro or VIP build is designed to carry the rest of the way.
Universal Cart didn’t create any of these requirements. It just made the cost of skipping them show up faster, in real time, on real traffic that’s already looking to buy.
Universal Cart decides how a shopper reaches your site; your own checkout still decides whether they buy. See how a ProXSites build handles that handoff at proxsites.com.
Sources
- Search Engine Journal, October 2026
- Think with Google, 2017 (2016 data)
- web.dev / Google, Core Web Vitals thresholds
- Baymard Institute, updated September 2025
Disclosure: ProXSites publishes this article and sells the website builds and hosting tiers it recommends, so it has a direct commercial interest in this advice. ProXSites is not affiliated with Google, Search Engine Journal or Baymard Institute, and reporting on Universal Cart here is sourced from their published material. Not legal or financial advice.
