Agentic Commerce Moves to Checkout as Google Arms Merchants for the 2026 Holidays
Google is betting that the 2026 holiday shopping season will be the first in which AI agents do real transaction work rather than merely answer product questions. The company published a set of agentic commerce updates on September 16 aimed at merchants preparing fourth-quarter campaigns, pairing a Google Shopping product post with a marketer-facing bundle from its Rethink 2026 event.
The pitch positions AI agents as a new layer sitting between merchants and shoppers. Discovery, comparison and, increasingly, the purchase itself move inside conversational surfaces rather than a traditional search results page or a retailer's own storefront.
This is not a one-month project. Google introduced the Universal Commerce Protocol (UCP) in January 2026 at NRF, co-developed with Shopify, Etsy, Wayfair, Target and Walmart. More than 20 partners endorsed the standard at launch, including Visa, Mastercard and Stripe, giving Google a payments and identity rail that spans most of US retail.
What Google Actually Shipped
The September updates extend features Google began rolling out to US shoppers on November 13. Agentic checkout now runs inside AI Mode and the Gemini app, so a user can complete a purchase without leaving the conversational interface. Merchants can offer discounts directly within AI Mode results, a change that turns promotional pricing into a machine-readable input rather than a banner a human has to notice.
Google has also deployed agents that call stores to check inventory, an unusual step that reaches into the least digitized layer of physical retail. The company says the US rollout will expand to other markets.
For merchants, the practical consequence is that product feeds stop being a display asset and start being a bidding and eligibility signal. If an agent can only recommend what it can parse, structured data quality becomes a revenue lever rather than a search optimization chore. Advertisers inherit a measurement problem at the same time: a conversion completed by an agent does not map cleanly onto click-based attribution, and Google has not published a reporting standard for it.
| Metric | Figure | Source |
|---|---|---|
| Agents' share of 2025 holiday retail sales | 20% ($262 billion) | Salesforce |
| Growth in generative AI referral traffic | +693% | Adobe |
| Projected agentic AI retail influence by 2030 | $3T–$5T | McKinsey |
| UCP launch partners | 20+ |
Why Google Is Racing to Own Agentic Commerce Checkout
The defensive logic is straightforward. Shopping queries are the commercial core of Google's search advertising business, and they are migrating toward conversational interfaces where the user asks a question and receives one answer instead of ten blue links. If that migration happens on someone else's assistant, Google loses the moment of intent that its ad auction monetizes.
Owning checkout keeps the transaction inside Google's surfaces and keeps the ad auction adjacent to the decision. The numbers explain the urgency. Salesforce data shows AI and agents influenced 20% of global retail sales in the 2025 holiday season, roughly $262 billion, though much of that use was customer service rather than assisted purchasing. Adobe measured a 693% surge in generative AI referral traffic during the same period.
The forward projections are larger. McKinsey estimates agentic AI could influence $3 trillion to $5 trillion in global retail by 2030, and Morgan Stanley projects that close to half of online shopping activity could involve agents.
The Open-Protocol Bet and Its Trade-offs
Google's choice to publish UCP as an open standard rather than a proprietary API is the most consequential decision in the bundle. Openness lowers adoption friction: Target and Walmart, competitors to Google's ad business in several respects, signed on because the protocol does not force them through a Google-controlled checkout.
The trade-off is control. An open protocol makes it easier for rival agents from OpenAI, Anthropic and Microsoft to transact against the same merchant catalog, which means Google can win the standard without winning the session. Amazon is pursuing the opposite strategy, keeping Rufus and its retail data inside a closed loop, and Walmart's Sparky follows a similar internal-first path.
Each model suits a different player. Merchants with strong first-party data and loyal customers gain from the closed approach because it protects margin and customer relationships. Merchants that depend on discovery, which is most small and mid-sized sellers, gain from open protocols because they lower the cost of being found. Google is wagering that the second group is larger, and that whoever defines the plumbing collects rent regardless of which assistant the shopper uses. That is the whole shape of the agentic commerce race.
The competitive clock adds pressure to that wager. Target had already gone live inside ChatGPT by the 2025 holiday season, and OpenAI has been building its own checkout path since. Amazon's Rufus and Walmart's Sparky evolved rapidly across the same period. Google's September bundle reads as an attempt to set terms before rival assistants reach enough scale to make merchant integrations a zero-sum choice. When the fourth quarter closes, that contest will be measured in merchant coverage rather than feature lists.
The Unresolved Parts: Returns, Delivery and Trust
Every announcement in this cycle leaves the same operational questions unanswered. When an agent completes a purchase, the delivery promise still belongs to the merchant, but the expectation was set by the assistant. A missed delivery window becomes a customer service failure for a retailer that never spoke to the buyer.
Returns follow the same logic. A purchase made inside AI Mode leaves no browsing session, no comparison page and no cart the shopper remembers, which complicates the dispute trail if an item arrives damaged or wrong. Visa, Mastercard and Stripe signed onto UCP, so payment authorization is covered. Liability for a bad outcome is not, and smaller sellers carry the most exposure because they absorb refunds without the logistics staff to contest a weak claim.
Consumer willingness to delegate payment credentials is the other open variable. Salesforce's finding that agents were used far more for customer service than for assisted purchasing during the 2025 holidays suggests the checkout habit is still forming. Google's bet is that a discount surfaced inside AI Mode is enough to close the gap between a recommendation and an authorization.
What Merchants Should Do Before Q4
Three operational moves follow from the announcements.
- Audit product feeds for machine readability, since agent eligibility depends on structured attributes, inventory accuracy and price consistency across surfaces.
- Decide how promotional discounts will appear in AI Mode, because a discount that only exists on a retailer's own site will not surface in an agent-mediated comparison.
- Instrument agent traffic separately, since AI referrals behave differently from paid clicks and blend into organic reporting by default.
There is a real cost here. Feeding a machine-readable catalog to an open protocol also feeds competitors' agents, and Google's own history shows that protocol owners capture disproportionate value over time. Merchants weighing UCP participation are trading short-term discoverability against long-term dependence on whichever agent layer wins.
Google's own advertising results hint at the playbook it expects. During the 2025 holiday season the company partnered with eight YouTube creators on an Uno card game takeover, which drove a 25% increase in searches for Uno. The lesson Google is selling to marketers is that agent-era discovery still starts with attention, and that attention still gets converted through Google surfaces.
Why this matters
The 2026 holiday season is the first real stress test of whether agentic commerce moves from pilot to habit. If agents convert at the rates the underlying data implies, the merchant question shifts from whether to participate to which protocol to back, and that answer will shape retail margins for years. The watch item for decision-makers is not the next feature announcement but whether UCP adoption holds through the fourth quarter under real transaction volume.
Sources
Boost your holiday sales with these agentic commerce updates
Google Shopping launches agentic checkout and more AI shopping tools
A new era of agentic commerce is here | Google Cloud Blog
Related Articles
- Google I/O 2026 Dates: Focus on Agentic AI
- Google Cloud Launches $750M Agentic AI Partner Fund
- Google I/O 2026 Schedule: Android 17 and the Rise of AI Agents
✔Human Verified
Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.