ChatGPT Shopping: when AI becomes your shopping engine

Last week, an e-commerce store owner shared a screenshot with me: her conversion rate from ChatGPT had just overtaken Google Shopping.

Since September 2025, ChatGPT is no longer just an assistant that answers questions: it’s a shopping engine that recommends products, compares prices, and completes transactions directly within the conversation.

The most accurate analogy is that of the digital personal shopper: where Google Shopping asks you to browse ten different pages, ChatGPT Shopping compresses the entire journey into three steps: ask, compare, buy.

This guide breaks down what this shift means in practice for e-commerce merchants, with the numbers, the technical mechanics, and the five actions to take right now.

ChatGPT Shopping: what just changed

700 million users consult ChatGPT every week to find products they want.

That’s the figure OpenAI highlighted at the launch of ChatGPT Shopping in September 2025: an already captive audience, already in active search mode, that merchants could reach directly from within the conversation.

The change isn’t cosmetic: it redefines the very structure of the buying journey.

Before ChatGPT Shopping, the typical journey looked like this: Google search, sorting through ads, visiting five websites, comparing prices, returning to Google, making a final decision.

With ChatGPT Shopping, the same journey fits in a single conversation: “What running shoes under $100 for wide feet?” The model recommends specific products, displays prices and availability, and offers a “Buy” button.

ChatGPT has shifted from trusted advisor to active seller: this isn’t a gradual evolution, it’s a structural break in the logic of online commerce.

Instant Checkout: buying without leaving the conversation

Instant Checkout is the core feature: it lets users complete a purchase directly inside the ChatGPT interface, without opening a single extra tab.

The process unfolds in a few clicks: the user asks a shopping question, ChatGPT displays organic, non-sponsored results, a “Buy” button appears on compatible products, the user confirms their details and payment method, and the order is sent to the merchant who handles delivery and customer service.

From day one, Etsy was a partner: millions of artisans suddenly had access to 700 million potential users without a single dollar spent on advertising.

Six months after launch, in March 2026, OpenAI announced a strategic pivot: the initial version of Instant Checkout “didn’t offer the level of flexibility” it had aimed for, and the company refocused ChatGPT on product discovery and comparison, leaving merchants in control of their own checkout.

This pivot doesn’t mark the end of AI commerce: it signals that ChatGPT’s real competitive advantage lies in the purchase intent it captures, not in the transaction itself.

Agentic Commerce Protocol: the open standard

Behind ChatGPT Shopping lies a technical infrastructure that most media cover poorly: the Agentic Commerce Protocol (ACP).

Co-developed by OpenAI and Stripe under Apache 2.0 license, the ACP is an open standard: any merchant can implement it with any compatible payment processor, with no proprietary lock-in.

The protocol rests on three founding pillars:

• Powerful: connects merchants to millions of AI users while preserving the direct customer relationship and “merchant of record” status
• Simple to adopt: if you already use Stripe, a single line of code activates agentic payments; with another processor, the delegated protocol allows integration without migration
• Flexible: works with all payment processors, all e-commerce platforms, all product and service types

The centerpiece is Stripe’s Shared Payment Token: a single-use token with a short lifespan and a defined maximum amount, which lets the merchant charge the buyer without ever accessing their raw card details.

Neither the merchant nor the AI agent sees real card data: everything remains tokenized within Stripe‘s secure system, with programmatic traceability for every use.

The ACP resolves the fundamental paradox of agentic commerce: letting an AI initiate a transaction while guaranteeing that the merchant retains full control over the customer relationship and PCI compliance.

From SEO to GEO: the new logic of e-commerce visibility

The question every e-commerce merchant should be asking is no longer “Am I on page one of Google?” but “Am I in ChatGPT’s answer?“

That’s the logic behind Generative Engine Optimization (GEO): instead of targeting a link algorithm, you structure your data and content to be cited and synthesized by language models.

What ChatGPT looks at when recommending a product

Here’s what most merchants don’t know: ChatGPT doesn’t crawl the web like Google to build its product carousel.

A Radyant analysis of 43,000 products in the ChatGPT carousel found that 83% of them match products in the top 40 of organic Google Shopping, with 60% coming directly from the top 10.

The conclusion is straightforward: ranking well on Google Shopping is the primary condition for existing in ChatGPT Shopping.

But that’s not enough.

ChatGPT then runs a second process: a web search for editorial context around those products, looking for reviews, comparisons, and mentions in authoritative sources.

A product that ranks in Google Shopping AND is regularly mentioned in quality editorial content holds a decisive edge over a product that relies solely on its feed.

The Product Feed becomes the primary authority source for your products: unlike Google Shopping, where external signals (links, mentions) matter as much as feed data, ChatGPT treats your feed as the primary reference.

Known ranking factors in the ChatGPT feed are:

• Freshness: the feed supports updates every 15 minutes, and stale data hurts overall visibility
• Consistency: feed data must match exactly what’s on the site, including return policies and displayed prices
• Availability: out-of-stock products degrade reliability signals across the entire merchant account
• Attribute richness: customer reviews, aggregated ratings, custom variants, videos, and 3D models create differentiation advantages the model can directly exploit

GEO + AI Shopping: the convergence

The convergence of GEO and AI Shopping creates a new visibility economy: LLM visitors are worth 4.4x more than organic visitors according to Semrush, because they arrive after having already conducted part of their thinking within the conversation.

The explanation comes down to one concept: intent compression, the phenomenon by which the user has already completed a large part of their decision-making in the conversation before landing on your site.

This logic connects directly to the AI search engine revolution underway: ChatGPT is no longer just an alternative to Google on shopping queries, it’s becoming a direct competitor on high purchase-intent segments.

GEO isn’t a separate discipline from SEO: it’s its logical extension when answers replace lists of links.

Real impact for e-commerce merchants

The reality for merchants is nuanced: native Instant Checkout remains US-only as of March 2026, but ChatGPT discovery traffic is already global.

In France, 31% of online shoppers already use generative AI in their purchase decision process, a figure that places the French market among the most advanced in Europe.

Carrefour has already launched a shopping experience integrated with ChatGPT: this is no longer a test, it’s a commercial reality deployed by a major player in French retail.

5 concrete actions to take right now

These five actions are ranked by impact and ease of execution:

1. Complete your Product Feed with rich attributes tailored to conversational queries
2. Prioritize Bing Merchant Center: ChatGPT pulls directly from Bing Shopping data for its product carousel
3. Develop product editorial content: independent reviews, comparisons, and specialist articles create the contextual signals ChatGPT analyzes in its second process
4. Synchronize inventory in real time: out-of-stock products degrade reliability signals across the entire merchant account
5. Start ACP integration via Stripe now, so you’re ready when European deployment accelerates

SMBs vs large retailers: comparative advantages

Counter-intuitively, specialized merchants have a structural advantage over generalists in ChatGPT Shopping.

An artisan selling handmade ceramics on Etsy has access to the same 700 million users as a national retailer, with a value proposition ChatGPT articulates with precision: “For a handcrafted stoneware mug, here’s the exact seller.”

Top-performing merchants are already recording 5% of their orders via AI discovery compared to 0.06% for the median: a clear premium for early adopters who invested in the quality of their product data.

This dynamic is part of the redefinition of the role of e-commerce sites in the face of AI agents: the storefront website is gradually giving way to the structured data feed as the primary interface with recommendation systems.

ChatGPT Shopping doesn’t reward catalog size, it rewards data precision: a rich, consistent, and fresh feed is worth more than a massive advertising budget on this channel.

Risks and limits you shouldn’t overlook

The conversion numbers are compelling, but several guardrails are needed before committing significant resources.

Absolute volume remains marginal: ChatGPT generates roughly 200x less e-commerce traffic than Google Organic and accounts for less than 0.2% of total traffic in the largest datasets.

Conversion data varies significantly by methodology: the 15.9% from Seer Interactive applies to specific verticals; Visibility Labs measures a more realistic rate of 1.81% vs 1.39% for organic across 94 sites.

The algorithmic dependency risk on ChatGPT is strategically serious: if OpenAI changes its eligibility criteria, your visibility can collapse without any direct recourse.

For European merchants, GDPR questions remain open: when your customer delegates a purchase to an AI agent, they share their consumption data with OpenAI, which requires a transparency clause in your privacy policies.

The European AI Act also imposes transparency obligations on providers like OpenAI for General Purpose AI systems, and merchants integrating ChatGPT Shopping must verify compliance.

Toward generalized agentic commerce

OpenAI‘s pivot on Instant Checkout in March 2026 is often read as a retreat: it’s a recalibration toward a more sustainable model.

The future of agentic commerce isn’t a “Buy” button inside ChatGPT: it’s an AI agent that knows your preferences, your budget, and your favorite brands, and autonomously initiates recurring purchases on the user’s instruction.

Semrush predicts that by 2027, AI channels will generate economic value comparable to traditional search channels at a global level.

Merchants preparing now, with a rich data feed, a solid Bing presence, a GEO editorial strategy, and an ACP integration in progress, will be in a decisive position when European deployment accelerates.

The first concrete step for your store: audit your Product Feed and check your presence on Bing Shopping.

The second: understand how to position your editorial content for generative engines by reading our GEO vs SEO guide.

FAQ

Is ChatGPT Shopping available in France?

As of March 2026, native Instant Checkout remains limited to US users, but discovery and recommendation traffic is already global: any merchant whose products appear in a Product Feed indexed by ChatGPT can receive referred visitors.

What is the Agentic Commerce Protocol (ACP)?

The ACP is an open technical standard under Apache 2.0 license, co-developed by OpenAI and Stripe, which defines how an AI agent can initiate a commercial transaction on behalf of a user while keeping the merchant as the “merchant of record“.

Why are ChatGPT conversion rates so high in some studies?

Intent compression explains these numbers: users receive a personalized recommendation within the ChatGPT conversation before arriving on your site, meaning their level of purchase certainty is structurally higher than that of an organic visitor.

What is Stripe’s Shared Payment Token?

It’s a single-use payment token created by the AI agent from the buyer’s saved payment method: it has a maximum amount and a short expiration time, ensuring it can only be used for a single transaction without exposing raw card data.

How does my Product Feed influence my visibility in ChatGPT?

83% of products in the ChatGPT carousel come from the top 40 of Google Shopping: your feed must be configured primarily for Google Merchant Center and Bing Merchant Center, with maximum freshness (updates every 15 minutes), full consistency between feed and site, and maximum attribute richness.

Can SMBs compete with large brands on ChatGPT Shopping?

Yes, and they can outperform them in their niche: ChatGPT Shopping values precision and specialization over catalog size, which gives a structural advantage to specialized merchants who can articulate a clear value proposition.

What are the GDPR risks of ChatGPT Shopping for European merchants?

When a customer delegates a purchase to an AI agent, they share their purchasing preferences and consumer behavior with OpenAI: merchants must add a transparency clause to their privacy policies and verify that data transfers to the United States comply with the EU-US Data Privacy Framework.

What’s the difference between ChatGPT Shopping and Google Shopping?

Google Shopping ranks links in a traditional interface; ChatGPT Shopping synthesizes a personalized recommendation within a conversation, with different ranking criteria (feed quality, editorial authority) and users who arrive with a more advanced purchase intent.

How do I prepare for ChatGPT Shopping’s arrival in Europe?

The five key actions are: enrich your Product Feed with all optional attributes, strengthen your presence on Bing Merchant Center, develop reference editorial content for your product categories, synchronize your inventory in real time, and start the technical ACP integration via Stripe.

Will ChatGPT Shopping replace Google Shopping?

In the short term, no: ChatGPT generates 200x less e-commerce traffic than Google Organic in absolute volume.

In the medium term (2027-2028), Semrush projects that AI channels will reach economic value comparable to traditional search channels, with gradually diverging visibility dynamics.