ChatGPT Shopping: How to Appear in AI Commerce Results

AI-driven traffic to U.S. retail sites surged 4,700% year-over-year by July 2025. That is not a typo. Adobe Analytics tracked the shift in real time as ChatGPT, Perplexity, and Google AI Mode quietly redirected how people discover products and services. Fifty-one percent of U.S. consumers used generative AI for shopping in 2025, up from 38% the year before. The channel that barely existed two years ago is now where buying decisions start.

And most businesses are completely invisible to it.

I have spent the last eighteen months building the infrastructure to make our company discoverable by AI shopping agents. Not theoretically. Operationally. We implemented every protocol, structured every data point, and built a scoring system that measures exactly how ready a business is for this shift. What I learned is that the gap between businesses that appear in AI commerce results and those that do not is not about marketing spend or brand awareness. It is about infrastructure.

What Is ChatGPT Is Now a Shopping Engine?

OpenAI did not announce this with a keynote. They rolled it out incrementally. ChatGPT now surfaces product recommendations, compares options, links to merchants, and facilitates purchases directly in the conversation. Users ask "What is the best project management tool for a 50-person agency?

OpenAI did not announce this with a keynote. They rolled it out incrementally. ChatGPT now surfaces product recommendations, compares options, links to merchants, and facilitates purchases directly in the conversation. Users ask "What is the best project management tool for a 50-person agency?" and ChatGPT does not just answer — it recommends specific products, compares pricing, and links to checkout flows.

This changes the entire discovery funnel. Traditional search required a user to type a query, scan results, click a link, navigate a website, and find what they need. AI commerce — sometimes called conversational commerce — compresses that into a single conversational exchange. The user asks. The AI shopping assistant answers. The transaction happens.

The businesses that appear in those answers capture the revenue. The businesses that do not appear get nothing. Not reduced traffic. Nothing. There is no page two of a ChatGPT response.

How ChatGPT Decides Who to Recommend?

This is where most businesses get it wrong. They assume AI recommendations work like search rankings — that it is about keywords, backlinks, and domain authority. It is not. ChatGPT and other AI shopping systems evaluate businesses on a fundamentally different set of signals.

This is where most businesses get it wrong. They assume AI recommendations work like search rankings — that it is about keywords, backlinks, and domain authority. It is not. ChatGPT and other AI shopping systems evaluate businesses on a fundamentally different set of signals.

Structured data carries the highest weight. When ChatGPT evaluates whether to recommend your business, it looks for machine-readable information it can parse and verify. JSON-LD markup describing your services, pricing, availability, and reviews. Schema.org vocabulary that tells the AI exactly what you offer, what it costs, and what credentials back it up. Product feed optimization — ensuring your catalog data is complete, consistent, and machine-readable — is no longer optional. Without this data, the AI has nothing reliable to work with. It will recommend a competitor that provides it.

Jorrit Steinz, CEO of ChannelEngine, put it directly: "If product data is not structured for machines, it will not surface where shopping now begins — and that means lost revenue before a buyer ever reaches your site." AI agents do not browse product pages. They evaluate structured data to make AI product recommendations. Missing or inconsistent attributes remove you from consideration entirely.

Beyond structured data, AI commerce systems evaluate three additional signal categories that determine which businesses receive AI product recommendations and which get ignored:

Why Most Businesses Score Under 30?

We built the ACRA framework — Agentic Commerce Readiness Assessment — specifically to measure this gap. Nine pillars covering structured data, protocol compliance, review presence, content authority, and five other dimensions that determine whether AI systems can find, evaluate, and recommend your business.

We built the ACRA framework — Agentic Commerce Readiness Assessment — specifically to measure this gap. Nine pillars covering structured data, protocol compliance, review presence, content authority, and five other dimensions that determine whether AI systems can find, evaluate, and recommend your business.

The results from our scans are consistent and stark. Out of the last 25 businesses we assessed, the pattern holds:

That is not cherry-picked. Those are real scans from our production database. Dan Martell runs a multimillion-dollar SaaS coaching business. Julian Goldie is an SEO agency owner with significant reach. MultiBank Group is a regulated financial institution. None of them score above 34.

The reason is consistent across every scan: no structured data beyond basic meta tags, no commerce protocol endpoints, insufficient review presence on platforms AI systems trust, and no machine-readable service catalog. They are operating on infrastructure built for Google circa 2015.

The Business Social Credit System

If this framework sounds familiar, it should. What is emerging in AI commerce is structurally identical to China's Social Credit System — but for businesses instead of citizens. In China, a composite score derived from financial behavior, social connections, purchasing patterns, and compliance history determines what a citizen can access. High score: preferential loans, faster service, better apartment listings. Low score: restricted travel, slower processing, reduced visibility in government systems. The scoring is algorithmic, the consequences are real, and most people do not fully understand how the score is calculated until it is too late.

AI commerce readiness works the same way. A composite score derived from structured data quality, protocol compliance, review presence, content authority, and trust signals determines what your business can access. High score: AI recommendations, conversational commerce visibility, agent-driven transactions. Low score: invisibility. Not a penalty. Not a demotion. Complete absence from the systems that increasingly determine who gets discovered.

The parallel is not accidental. Both systems are algorithmic gatekeepers that evaluate entities across multiple dimensions and produce a single trust verdict. Both operate continuously — your score changes as your signals change. Both reward infrastructure investment and punish neglect. And both are fundamentally opaque to the entities being scored, which is exactly why we built the ACRA scanner — to make the invisible score visible, to show businesses exactly where they stand before the consequences become irreversible.

The difference is that China's system is mandated by the state. The AI commerce scoring system is mandated by the market. No government required you to implement structured data or deploy protocol endpoints. But the AI systems that now mediate $385 billion in commerce decisions will ignore you if you do not. The mandate is economic, not political — and arguably more consequential for your revenue.

What Is The Three Protocols That Make You Visible?

AI commerce runs on protocols, not pages. Three standards define how AI agents discover, transact with, and trust businesses. If you do not implement them, you are not in the conversation.

AI commerce runs on protocols, not pages. Three standards define how AI agents discover, transact with, and trust businesses. If you do not implement them, you are not in the conversation.

UCP: How AI Agents Find You

The Universal Commerce Protocol, published by Google, is the discovery layer. A JSON manifest at /.well-known/ucp/manifest.json tells every AI shopping agent what your business sells, how to transact, and which transport protocols you support. Without it, an AI agent visiting your domain finds nothing actionable and moves to a competitor that has one.

This is binary. You either have a UCP manifest or you do not exist to agent-driven commerce. There is no partial credit.

ACP: How AI Agents Buy From You

The Agentic Commerce Protocol, published by OpenAI, governs the checkout conversation. If UCP is how agents find you, ACP is how they buy from you. An agent operating under ACP negotiates terms, applies user preferences, and completes payment — all in a single automated exchange that takes seconds.

The businesses that implement ACP see conversion rates from AI-driven traffic that make traditional checkout flows look primitive. The businesses that do not implement ACP watch agents bounce at the checkout wall.

AP2: How AI Agents Trust You

The Agent Payments Protocol provides the cryptographic trust layer. Signed mandates authorize exactly what an agent can spend, with whom, and under what constraints. Without AP2, there is no agentic commerce. Discovery and checkout are useful only when the agent can complete a transaction with cryptographic certainty.

Together, these three protocols form the complete stack: UCP for discovery, ACP for checkout, AP2 for payment trust. We implemented all three at Adam Silva Consulting. Our ACRA score reflects it.

What We Built and Why Traditional Agencies Cannot?

When we decided to make our own business agentic-ready, we discovered something the market had not articulated yet: the gap between knowing about these protocols and actually implementing them is enormous. It requires structured data expertise, payment infrastructure, cryptographic signing, and deep understanding of how AI systems parse and evaluate business information.

When we decided to make our own business agentic-ready, we discovered something the market had not articulated yet: the gap between knowing about these protocols and actually implementing them is enormous. It requires structured data expertise, payment infrastructure, cryptographic signing, and deep understanding of how AI systems parse and evaluate business information.

Traditional digital agencies sell SEO, PPC, and social media management. They optimize for Google's legacy search algorithm — keywords, backlinks, meta descriptions. None of that infrastructure translates to AI commerce. A business that ranks first on Google for a target keyword can score 10 out of 100 on an ACRA assessment because the two systems evaluate completely different signals.

We built the ACRA scanner because no tool existed to measure what matters. Nine pillars, each weighted by the signal's importance to AI agent decision-making:

1. Structured Data Quality — Schema.org markup depth and accuracy
2. Protocol Endpoints — UCP, ACP, AP2 manifest presence and validity
3. Review Presence — Cross-platform review profiles (GBP, G2, Trustpilot, Capterra)
4. Content Authority — Expert content depth, E-E-A-T signals, defined terms
5. Technical SEO Foundation — Server-side rendering, Core Web Vitals, sitemap health
6. Social Proof — Platform presence, engagement signals, brand consistency
7. Security and Trust — SSL, privacy policy, terms of service, trust badges
8. Schema Completeness — Organization, LocalBusiness, Service, Product schema types
9. AI Citation Readiness — Speakable markup, FAQ schema, answer-format content

The score tells you exactly where you stand. More importantly, it tells you what to fix first. Every point on the ACRA scale maps to a specific infrastructure gap with a specific revenue impact.

What You Can Do Right Now?

I am not going to pretend this is simple. Full protocol implementation takes weeks of specialized engineering. But there are steps that move the needle immediately.

I am not going to pretend this is simple. Full protocol implementation takes weeks of specialized engineering. But there are steps that move the needle immediately.

First, get your score. Run an ACRA assessment on your domain. The scan takes seconds and shows you exactly which of the nine pillars are failing. Most businesses discover they are missing 6 or 7 out of 9. That is the starting point — not panic, but precision.

Second, implement structured data. This is the highest-leverage single change. Add JSON-LD markup to every page that describes your business, services, pricing, and team using schema.org vocabulary. Organization, Service, LocalBusiness, Person, FAQ — each schema type gives AI systems another data point to work with.

Third, build your review presence. AI systems cross-reference reviews before recommending businesses. Google Business Profile, G2, Trustpilot, and Capterra are the platforms that matter most. Five reviews on each platform is the minimum threshold our ACRA scanner evaluates.

Fourth, publish expert content AI can cite. ChatGPT and other AI systems prefer sources with depth, specificity, and verifiable expertise. This is where GEO optimization becomes critical — structuring your content so generative engines can extract and cite it. Articles with defined terms, FAQ schema, and speakable markup are dramatically more likely to be cited than generic blog posts. The Authority Flywheel compounds these signals over time.

Fifth, deploy protocol endpoints. This requires engineering, but the infrastructure is well-defined. A UCP manifest, an ACP configuration, and AP2 mandate verification — deployed at standardized .well-known URLs on your domain. Our UCP Implementation service handles the full stack.

What Is The Window Is Closing?

McKinsey estimates generative AI could unlock $240 billion to $390 billion in value for retailers. Autonomous agents could influence $385 billion in U. S. e-commerce by 2030. Ninety-seven percent of retailers plan to increase AI spending in the next fiscal year.

McKinsey estimates generative AI could unlock $240 billion to $390 billion in value for retailers. Autonomous agents could influence $385 billion in U.S. e-commerce by 2030. Ninety-seven percent of retailers plan to increase AI spending in the next fiscal year. And here is the number that should make every executive pay attention: traffic from ChatGPT and AI shopping agents converts at 25 times the rate of traditional search traffic. These are not window shoppers. These are buyers with intent, filtered and qualified by AI before they ever reach your checkout.

The businesses that build AI commerce infrastructure now will compound their advantage. The ones that wait will face a market where competitors have established protocol presence, accumulated review signals, and built the content authority that AI systems trust. Catching up from zero while competitors compound is not a strategic position anyone wants.

We saw this shift coming. We built the infrastructure. We created the scoring system. We documented the protocols. And we made it all available because the market needs it — not in 2028, but right now.

The question is not whether AI commerce is happening. The 4,700% traffic surge answered that. The question is whether your business is visible to the systems that are making purchasing decisions on behalf of millions of consumers. Right now, for most businesses, the answer is no.

That is fixable. But not by waiting.

Last Fact-Checked & Metric-Verified: March 2026 · Sources cited inline with publication year