Infographic: Embedded Checkout and Direct Offers: UCPs Two Transaction Models

Embedded Checkout & Direct Offers: UCP’s Two Transaction Models Explained

🎧 Listen to this article
Last Updated: November 10, 2026 • Autonomous Agentic Commerce Spec v2026-08-25

Production Validated

The Universal Commerce Protocol defines two distinct transaction models that determine how a purchase actually happens when an AI agent acts on behalf of a shopper: Native Checkout and Embedded Checkout. Understanding the difference between these models — and choosing the right one for your business — is one of the most consequential technical decisions a merchant makes during UCP implementation.

Alongside these checkout models, Google has introduced Direct Offers, a new Google Ads capability that lets advertisers present exclusive discounts to shoppers within AI Mode. Together, these three mechanisms define the transaction layer of agentic commerce.

Definition: UCP’s transaction models determine the boundary between the AI agent’s interface and the merchant’s checkout system. Native Checkout processes the entire transaction within the AI surface using UCP-standardized API calls. Embedded Checkout renders the merchant’s own checkout UI components within the AI surface, preserving custom checkout experiences.
Architecture Dimension Native Checkout (Protocol Rail) Embedded Checkout (In-Canvas Iframe) Direct Offers (AI Mode Promotion) PCI-DSS Scope Optimal Merchant Fit
Execution Plane 100% Machine-to-Machine REST/MCP API transport Sandboxed in-canvas web component or secure browser frame Merchant-sponsored structured discount injected into conversational stream Zero Card Data Touches Agent High-frequency replenishment & consumables
Consumer Handoff Friction Zero redirect; complete purchase inside conversational turn Micro-modal pops up within assistant UI for payment entry One-tap activation pill embedded directly beneath recommended product SAQ A Compliant Custom configured apparel & subscription plans
Settlement & Token Rail Network Tokenization v3 + FIDO AP2 Delegated Mandate Direct merchant payment gateway session (Stripe / Google Pay) Automated dynamic voucher code + cryptographic pricing claim PSP Isolated Enclave Flash sales & algorithmic ad conversions
Brand Merchandising Control Deterministic JSON schema; agent formats presentation 100% merchant visual CSS styling control inside iframe canvas Merchant defines exact badge copy, discount tiers & expiration countdown No Agent Tampering Risk Luxury & design-forward retail brands
Conversion Lift Over External URL +38% higher repeat order conversion +24% lift over external checkout links +47% conversion on high-intent shopping queries Cryptographic Audit Proof All UCP-certified merchants

Native Checkout: Full Protocol-Layer Transactions

When a merchant configures the default REST API transport binding in their UCP discovery endpoint, they are telling AI agents that their store can process a complete transaction entirely through UCP’s standardized API calls — no redirect to the merchant’s website, no rendering of the merchant’s checkout pages, no browser interaction required.

In a Native Checkout flow, the AI agent handles the entire user-facing experience. The shopper expresses intent conversationally (“buy this one”), and the agent initiates a checkout session by sending a POST request to the merchant’s checkout API with the selected product, quantity, shipping address, and payment credentials. The merchant’s backend processes the order — validating inventory, calculating taxes and shipping, authorizing payment — and returns an order confirmation to the agent, which presents it to the shopper.

From the shopper’s perspective, the entire transaction happens within the conversation. There is no page load, no form to fill out, no redirect. From the merchant’s perspective, the order arrives through their existing order management system as if it came from any other sales channel.

When to Choose Native Checkout

Native Checkout is the optimal model when your products have straightforward purchase flows — standard items with clear pricing, simple shipping options, and no mandatory customization steps. It delivers the lowest-friction shopping experience and the highest conversion rates because there are zero interface transitions that could cause drop-off.

Retailers selling consumer goods, electronics, books, beauty products, apparel in standard sizes, and similar catalog items are ideal candidates for Native Checkout. The model works best when the shopper’s primary decision is “which product” rather than “how to configure this product.”

Embedded Checkout: Merchant-Controlled UI Within AI Surfaces

Embedded Checkout — signaled by including the embedded transport in the services definition of the discovery document — takes a different approach. Instead of the AI agent handling the full transaction, the merchant’s own checkout interface renders within the AI surface as an embedded component. The shopper interacts with the merchant’s checkout UI — which can include custom forms, configuration options, delivery schedulers, and branded experiences — without leaving the AI conversation.

According to Google’s developer documentation, Embedded Checkout is an optional path available to specific, approved merchants with bespoke branding or complex checkout flows. This model is designed for merchants whose checkout process involves decisions that require visual interfaces or interactive elements that cannot be easily expressed through conversational text. A furniture retailer that requires customers to select a specific delivery date and time slot, a custom jewelry store that needs shoppers to choose engraving options, or a travel company with complex booking configurations would all benefit from Embedded Checkout.

When to Choose Embedded Checkout

Choose Embedded Checkout when your products require configuration or customization during the purchase process, your checkout flow includes legally required disclosures or consent steps that must be presented in a specific format, you want to maintain complete visual control over the checkout experience including branding, or your transaction involves complex multi-step decisions like delivery scheduling, assembly options, or service tier selection.

The trade-off is slightly higher friction — the shopper transitions from a conversational interface to a visual checkout component. However, for products that genuinely require these interactive elements, the alternative of forcing everything into a text-based conversation would actually increase friction and reduce completion rates.

Direct Offers: Exclusive Deals in AI Mode

Direct Offers is a separate but complementary capability announced alongside UCP’s broader rollout. Unlike Native and Embedded Checkout — which are UCP protocol features available to any implementing merchant — Direct Offers is a Google Ads product that lets advertisers present exclusive discounts to shoppers who are actively researching and ready to buy within AI Mode in Google Search.

When a shopper is conversing with AI Mode about a product category, advertisers participating in the Direct Offers pilot can present time-limited, exclusive promotions — for example, “20% off this item, available only through this AI Mode conversation.” The offer is presented as part of the AI’s response, alongside organic product results.

Direct Offers bridges the gap between advertising and agentic commerce. Traditional display and search ads drive traffic to a merchant’s website. Direct Offers drive immediate transactions within the AI conversation, combining promotional incentives with the zero-friction checkout experience enabled by UCP.

How Direct Offers Work With UCP Checkout

When a shopper accepts a Direct Offer, the transaction flows through the same UCP checkout infrastructure — either Native or Embedded — as any other AI-initiated purchase. The promotional pricing is applied at the checkout session level, meaning the merchant’s backend sees the discounted price and processes the order accordingly. The AI agent handles the presentation of the offer and the shopper’s acceptance, then initiates the standard UCP checkout flow.

Technical Implementation: Declaring Your Checkout Model

Your checkout model is declared in your /.well-known/ucp discovery document within the services object. When the embedded transport binding is present alongside the rest transport, the merchant supports both integration paths. The default REST API path handles standard checkout, while the embedded path provides an iframe-based UI for products that require configuration.

When both transport bindings are present, the AI surface can determine which path to use based on the transaction context. This means a single merchant can offer a seamless API-based checkout experience for standard catalog items while providing an iframe-based branded checkout for configurable or complex products.

The native_commerce Attribute

For products to be eligible for UCP-powered checkout — whether Native or Embedded — they must have the native_commerce attribute set in their Google Merchant Center product feed. This attribute serves as the opt-in signal that tells Google’s AI surfaces these products can be purchased through the agentic checkout flow.

Products without the native_commerce attribute may still appear in AI search results and product recommendations, but shoppers will be redirected to the merchant’s website to complete the purchase rather than being offered an in-conversation checkout experience. Setting this attribute is therefore a prerequisite for both Native and Embedded Checkout.

Choosing the Right Model: A Decision Framework

The decision between Native and Embedded Checkout is not about which is “better” — it is about which model fits your products and your customers’ purchase journey. Start by asking: can a shopper make a fully informed purchase decision for this product based on the information an AI agent can present in a conversation? If yes, Native Checkout delivers the best experience. If the shopper needs to interact with visual configuration tools, select from complex options, or view specific legal disclosures, Embedded Checkout is the right choice.

Many merchants will use both. A sporting goods retailer might offer Native Checkout for standard items like running shoes and protein powder while using Embedded Checkout for custom team jerseys that require name and number selection. The UCP discovery document supports this mixed approach natively.

UCP Transaction Models & Checkout Architecture FAQ

When should an enterprise brand choose Native Checkout over Embedded Checkout?

Enterprise brands should deploy Native Checkout for standardized SKUs, recurring consumables, and repeat purchases where minimizing purchase friction is paramount. Because Native Checkout executes entirely over machine-to-machine APIs without rendering a browser canvas, conversion rates increase by up to 38%. Conversely, Embedded Checkout is ideal when legal disclosures, complex bundle configuration, or custom gift-wrapping requires the merchant’s proprietary UI.

How does Google AI Mode verify merchant eligibility for Direct Offers?

Google AI Mode validates Direct Offers by querying the merchant’s /.well-known/ucp manifest and matching it against active Google Merchant Center promotions feeds. The offer must include a cryptographically verifiable discount token and an active inventory reservation schema. Merchants with verified UCP conformance badges receive algorithmic priority in high-intent conversational product cards.

Does Embedded Checkout require merchants to maintain Level 1 PCI-DSS compliance?

No. Embedded Checkout leverages sandboxed WebMCP frames and hosted fields provided by certified payment service providers (such as Stripe Elements or Google Pay API). Because consumer payment credentials bypass both the AI assistant and the merchant’s application server, merchants qualify for simplified PCI Self-Assessment Questionnaire A (SAQ A), eliminating costly security infrastructure audits.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *