Why Your Products Don't Appear in ChatGPT Shopping

Most AEO tools measure whether AI mentions your brand. That is not the same as getting your products recommended for purchase. Your product feed must satisfy the technical requirements of OpenAI's Agentic Commerce Protocol (ACP) and Google's Universal Commerce Protocol (UCP) before AI agents can recommend or transact your products at all. Brand mention tracking cannot diagnose or fix a non-compliant feed. The real bottleneck for eCommerce merchants is protocol compliance: that requires SKU-level feed auditing, a different product category entirely.

Last updated: 28 September 2026. Checklists re-verified against OpenAI's current product feed spec and Google's UCP Merchant Center guide; outdated field requirements corrected.

Key Takeaways

• 50+ AEO tools now compete to track AI brand mentions (AEOsome market audit, 2026). None audit eCommerce protocol compliance.
• A one-year academic study found affiliate links converted 86% better than ChatGPT-referred traffic for eCommerce (Digiday, 2026), but that conversion picture is shifting fast as AI shopping volume scales, which is exactly why eligibility, not conversion tactics, is the durable lever.
• ACP and UCP are hard technical gates. A non-compliant product feed means AI agents skip your products regardless of how much content you publish.
• The failures most merchants don't know about: unstable variant IDs, malformed price or availability values (which reject the row outright), missing GTINs that weaken catalog matching, and stale stock status.
• Fixing this requires a Monitor → Audit → Optimize → Verify loop at the SKU level, not a dashboard that counts brand mentions.

In this article:

• The market-monitoring misunderstanding
• What actually happens when a shopper uses ChatGPT to buy something
• Shopify's Agentic Plan gets you enrolled. Enrollment is not the same as appearing.
• The market audit: what 50+ tools actually do
• What protocol compliance actually requires: a checklist
• The four-step operational fix
• Why this matters now
• FAQ

The market-monitoring misunderstanding

In the past 18 months, over 50 tools have launched to help brands "optimize for AI search." The pitch is consistent: track how often ChatGPT, Perplexity, Gemini, and Copilot mention your brand. Improve your mention rate. Win AI visibility.

It is a compelling pitch. And for B2B SaaS companies, professional services firms, and publishers, it is roughly the right problem to solve.

For eCommerce brands, it is the wrong problem entirely.

Here is the gap the market has not yet absorbed: AI brand mentions and AI product recommendations are two different things, powered by two different systems, with two different sets of requirements.

When ChatGPT mentions your brand in a paragraph about industry trends, that is brand visibility, the thing those 50+ tools track. When ChatGPT surfaces your specific products in a shopping carousel that a user can click and purchase, that is product eligibility, governed entirely by technical protocol compliance.

No mention-tracking tool can fix a compliance failure. A one-year academic study of eCommerce sites found that affiliate links converted 86% better than ChatGPT-referred traffic (Digiday, 2026), but that gap is closing as AI shopping volume scales. The more durable problem, regardless of which way conversion rates trend, is that a non-compliant product never gets shown to a shopper in the first place.

What actually happens when a shopper uses ChatGPT to buy something

To understand why tracking is the wrong starting point, it helps to understand what actually happens when a shopper types "best waterproof hiking boots under $200" into ChatGPT.

The AI does not search the web for your product pages. It does not read your blog posts. It queries a real-time product feed that merchants have registered through OpenAI's Agentic Commerce Protocol, a structured data API that works like a modern commerce backbone, not a search index.

For your product to appear in that query result, each row of your feed has to pass OpenAI's product feed spec at the variant level:

• Stable variant IDs (required): each individual SKU (size 9, color black) needs a stable ID that is unique in the feed and never reused for a different item
• Valid availability (required): one of in_stock, out_of_stock, pre_order, backorder, or unknown. Per the spec, omitted, empty, or unrecognized values reject the row
• Correctly formatted price: written as amount CURRENCY, for example 79.99 USD, with a three-letter ISO 4217 currency code
• GTINs (optional, but they matter): the spec says to omit a GTIN if none is assigned and never to invent one. It is still the key engines use to match your product across sellers, and Google limits visibility for products that omit a manufacturer-assigned GTIN
• Policy URLs (optional): return policy links must be valid HTTP(S) URLs or the parser drops them
• Feed freshness: the spec asks you to submit current stock status with each row and to update the feed when a sale starts or ends. A stale feed means an agent can recommend something you can't ship

If a required field is missing or malformed, that row is rejected. If a recommended field like GTIN is missing, the product is harder to match and easier to outrank. Either way, the fix lives in your feed, not in your product descriptions or in how often your brand is cited in blog posts.

On Google's side, the Universal Commerce Protocol adds another layer. According to Google's developer documentation, businesses publish the services and capabilities they support (checkout, identity linking, order management) in a JSON manifest at /.well-known/ucp, which agents read to discover what your store can do. A store with no manifest gives Google's agentic checkout nothing to connect to, even if it ranks on Google Search.

Shopify's Agentic Plan gets you enrolled. Enrollment is not the same as appearing.

Credit where it is due: Shopify has moved faster on agentic commerce than almost any platform in the market. The Agentic Plan auto-enrolls merchants into Shopify Catalog, a real-time product feed layer that connects directly to ChatGPT, Copilot, Google AI, and Gemini without any integration work from the merchant. Shopify also deploys specialized LLMs to enrich product data, filling in attribute gaps that would otherwise reduce a product's visibility. Brands like Gymshark, Everlane, Fenty Beauty, and Monos are already live. Shopify merchants who signed up for the Agentic Plan are, in a meaningful sense, already in the game.

But enrollment and appearing are two different things.

Shopify Catalog gets your products into the discovery pool. What determines whether they surface in a specific shopping query (and whether they rank above competitors) is the quality and completeness of your variant-level data. Shopify's LLM enrichment helps, but it cannot fabricate data that does not exist in your catalog, and it cannot fix structural compliance gaps that sit below the product description layer.

Here is what still falls through:

GTINs for private-label and custom products. Shopify's LLMs can rewrite a product title. They cannot generate a valid barcode for a product that has never been assigned one. Merchants who sell custom, white-label, or handmade products (a significant portion of the Shopify merchant base) still carry this gap. OpenAI's feed spec lets you omit a GTIN, so the row isn't rejected. But the product loses the identifier engines use to match it across sellers and sources, and Google Merchant Center limits visibility when a manufacturer-assigned GTIN is left out. The product can be eligible and still lose to a competitor whose record is easier to match. A brand profile assembled from conflicting sources can keep your catalog out of the running, and fixing entity consistency covers the five-step correction across every source AI reads.

Feed freshness for high-velocity SKUs. Shopify Catalog syncs automatically, but the frequency is not always sufficient for fast-moving inventory. OpenAI's spec asks for current stock status on every row and a feed update whenever a sale starts or ends. An agent that recommends an item you've sold out of is a failed purchase for the shopper and a wasted slot for you. Merchants selling hundreds of units a day of a small SKU set need to check whether their sync cadence matches their sell-through rate. This is operational, not structural, but it degrades over time without active monitoring.

The Google channel is a separate setup entirely. Shopify's Agentic Plan handles the ChatGPT and Copilot channels. Google's Universal Commerce Protocol is a different system. Google's UCP Merchant Center guide requires an account in good standing with approved free-listing products, return policies and at least one customer-support contact defined in Merchant Center, and a checkout-eligibility attribute on each product. Products that leave that attribute unset default to not eligible. Several categories are excluded outright, including subscriptions, pre-orders, personalized goods, and used or refurbished items. None of this is automated by Shopify's Agentic Plan. It requires deliberate setup on the Google side.

Variant-level attribute completeness. Shopify Catalog and LLM enrichment improve descriptions. They do not guarantee that the decision-making attributes AI agents match against (material, dimensions, fit, use case, care instructions) are present, consistent, and correctly mapped at the individual variant level. Our own data shows why that matters. In AEOsome's study of 8,520 ChatGPT shopping queries (Women's Fashion, US market, 9 intent stages), queries naming 5 product attributes triggered a product carousel 87.3% of the time, compared with 58.1% for queries naming none. Shoppers ask in attributes, and the agent can only match products whose variant data actually carries them.

The honest picture: Shopify has done the hard work of building the pipe. The question is what flows through it. Enrollment gets you in the door. Compliance and data quality determine whether you appear, and where.

The market audit: what 50+ tools actually do

A review of the AEO tools market (50+ tools, from $19/mo freemium to $150K/yr enterprise) reveals a consistent pattern: nearly every tool is a brand visibility tracker. Method: AEOsome reviewed each tool's publicly documented feature set and pricing page ahead of this post's original April 2026 publication. Tools were counted as eCommerce-targeted only if they positioned themselves for online stores, and as SKU-level only if they documented per-variant reporting.

The typical feature set:

• Input a brand name or a set of prompts
• Query ChatGPT, Perplexity, Gemini, and Copilot
• Report back: how often does your brand appear? What sentiment? What competitors appear alongside you?

This is useful for brand teams. It answers "are we showing up in the conversation?" It does not answer "are our products eligible to appear in shopping carousels?" and it cannot answer "what is blocking our SKUs from appearing in purchase-intent queries?"

Of the 50+ tools surveyed, the breakdown is stark:

The two eCommerce-focused tools that exist track brand-level visibility in ChatGPT Shopping and Amazon Rufus respectively. Neither addresses the feed compliance layer. Neither operates at SKU/variant level. Neither is aware of ACP or UCP as a compliance framework.

This is not a criticism of those tools; they are doing the right thing for the problem they have defined. The gap is that no tool in the market has yet addressed the eligibility layer that actually determines whether eCommerce products appear.

What protocol compliance actually requires: a checklist

If you are a Shopify merchant trying to assess your own ACP/UCP readiness today, start here.

ACP compliance checklist

• Every variant has a stable, persistent variant ID (does not change when you edit the product, never reused for a different item)
• Every row has a valid availability value: in_stock, out_of_stock, pre_order, backorder, or unknown
• Prices are written as amount CURRENCY (for example 79.99 USD)
• Every variant that has a manufacturer-assigned GTIN carries it. Where none exists, it is omitted, never invented
• Return policy URL, if supplied, is a valid public HTTP(S) link, not behind a login
• Feed updates whenever stock status changes or a sale starts or ends
• Shopify Agentic Plan is activated (handles feed submission to ChatGPT and Copilot automatically, no manual submission required for Shopify merchants)
• Variant titles are descriptive enough for an AI agent to differentiate ("Blue / Size 8" is insufficient; "Cobalt Blue / Women's US Size 8 / Wide Width" is better)

UCP compliance checklist

• /.well-known/ucp manifest is published at your domain
• Manifest declares the capabilities you support (checkout at minimum; identity linking and order management if you offer them)
• Google Merchant Center account is in good standing, with approved products for free listings
• Return policy (return cost, return window, link to full policy) is defined in Merchant Center
• At least one customer-support contact (website, email, or phone) is set in Merchant Center
• Checkout-eligibility attribute is set on every product you want purchasable. Missing means not eligible
• Product IDs in your feed match the IDs your checkout API expects
• You have confirmed none of your products fall in an excluded category (subscriptions, pre-orders, personalized, used or refurbished, age-restricted, and others)

If you cannot check every box in both lists, you have eligibility gaps, not content gaps.

The four-step operational fix

Protocol compliance is not a one-time project. Feeds drift. Inventory changes. New SKUs launch. Policy pages move. The compliance status of a 1,000-SKU catalog is a dynamic number that degrades without active maintenance.

The operational approach that works is a closed-loop system: Monitor what appears, Audit what is blocking what does not, Optimize the specific fields and feed structures causing failures, then Verify that changes produced the intended result.

Monitor: Run systematic shopping queries across purchase-intent prompts for your categories. Measure which SKUs appear, at what rank, with what share of voice. Build a weekly baseline.

Audit: For every SKU not appearing, diagnose the root cause. Is it a compliance failure (missing field)? A competitiveness gap (attribute completeness below winning competitors)? A query-intent mismatch (your product data does not semantically align with what the query asks)?

Optimize: Prioritize fixes by compliance status first, a non-eligible product cannot be ranked, so ranking improvements are irrelevant until eligibility is resolved. Fix GTINs, policy URLs, inventory freshness, and feed structure before touching descriptions or adding attributes.

Verify: Re-run the same query set 48–72 hours after changes. Measure the delta: SKU Visibility Rate, Average Shopping Rank, Share of Voice. This produces the before/after data your leadership needs to justify the work.

This loop runs weekly. The target AEOsome sets for it is 90%+ of a catalog's SKUs passing every ACP check within six weekly cycles. That is a planning target, not a guaranteed outcome. How fast a catalog gets there depends mostly on how many variants lack identifiers you can actually source, because a missing GTIN can't be fixed by editing a field; someone has to find the real barcode.

Why this matters now

April 16, 2026 was a meaningful date for eCommerce. OpenAI relaunched ACP at scale. ChatGPT had passed 900 million weekly active users by OpenAI's February 2026 disclosure, and product shopping carousels are now available to that base. 1 million+ Shopify merchants are being onboarded to the platform via Shopify's Agentic Plan. ACP has been open-sourced. Google's Universal Commerce Protocol is live for Shopify, Etsy, Wayfair, Target, and Walmart merchants, with Meta also moving toward agentic commerce for Facebook and Instagram.

The infrastructure is live, and so is the audience. OpenAI says shoppers already ask ChatGPT 50 million shopping-related queries a day (Digiday, December 2025). 42% of US consumers used at least one AI tool to shop in the past month (NIQ, May 2026). For how that usage maps onto the buying journey, see our breakdown of the AI shopping customer journey.

The gap is eligibility. Most merchant catalogs were not built to satisfy protocol requirements. The brands that audit and fix their compliance in the next 6 months will own a disproportionate share of AI shopping traffic before the broader market catches up.

Brand mention tracking is the wrong tool for this job. The right tool is one that operates at the SKU level, speaks the language of ACP and UCP, and tells you specifically which products are blocked and why.

FAQ

Does publishing more content help my products appear in ChatGPT shopping carousels?

Content improves your brand's presence in AI-generated answers. It does not make a product eligible for a shopping carousel. Carousels are built from product feeds, and under OpenAI's product feed spec a row with a missing or unrecognized availability value is rejected outright, however much content surrounds it.

My brand shows up in ChatGPT conversations. Why aren't my products appearing in shopping results?

Brand mentions and product carousels come from different systems. Mentions come from training data and web citations; carousels come from merchant product feeds. The two are independent. In AEOsome's study of 8,520 ChatGPT shopping queries, only 59.5% of shopping queries triggered a product carousel at all, so being mentioned in an answer is not the same as having products shown in it.

What is the most common reason Shopify products fail ACP compliance?

Malformed required fields and unstable variant IDs are what get rows rejected. OpenAI's feed spec rejects any row with an omitted or unrecognized availability value and requires a stable ID per variant that is never reused. Missing GTINs are the most common quieter gap: the spec lets you omit them, but products without one are harder for engines to match across sellers.

How long does it take to get a catalog ACP-compliant?

AEOsome's working target is 90%+ of a catalog's SKUs passing every ACP check within six weekly audit-and-fix cycles. That is a planning target, not a guarantee. The biggest variable is how many variants lack a real GTIN, because the spec forbids inventing one; someone has to source the actual barcode.

Is UCP compliance required if I am already ACP-compliant?

Yes, if you want to sell through Google. ACP covers ChatGPT; UCP covers Google's agentic checkout in AI Mode and Gemini. Google's UCP Merchant Center guide adds its own requirements: an account in good standing, return policy and customer-support contact defined in Merchant Center, and a checkout-eligibility attribute on each product, which defaults to not eligible when missing.

Do I need a developer to fix ACP compliance issues?

Some fixes can be done in Shopify by a non-technical team member, such as entering real GTINs and updating policy URLs. Feed structure issues, variant ID mapping, and publishing a UCP manifest at /.well-known/ucp (Google Developers) usually need a developer or a tool that handles the technical layer. Audit first: without a SKU-level compliance report you can't tell which kind of gap you have.

Is your Shopify catalog ACP-compliant? Most merchants don't know until they check. AEOsome's free audit scans your product feed for ACP and UCP compliance gaps (at the SKU level) and shows you exactly what is blocking your products from appearing in ChatGPT, Gemini, and Google AI shopping surfaces. → Get your free audit here

About the author. Vijaya Kumar Channalli is a serial SaaS founder and the founder of AEOsome (Tartu, Estonia), which helps eCommerce brands get their products recommended when shoppers ask AI what to buy. He built AEOsome's SKU-level ACP/UCP compliance auditing methodology and writes about AI shopping visibility, product feeds, and how Answer Engine Optimization differs from SEO. More at vijay.cc. Read how we write, source and correct posts in our editorial policy.

Sources: AEOsome market audit of 50+ AEO/GEO tools (April 2026, public feature and pricing pages); AEOsome, 8,520 ChatGPT shopping queries study (2026); OpenAI, Agentic Commerce Protocol documentation (2026); OpenAI, product feed specification (accessed September 2026); Google, UCP Merchant Center guide (accessed September 2026); Google Developers, Under the hood: Universal Commerce Protocol (2026); Search Engine Land, ChatGPT 900 million weekly active users (February 2026); Digiday, OpenAI's ChatGPT shopping research tool (December 2025); NIQ, 42% of consumers now use AI tools to shop (May 2026); Google, Universal Commerce Protocol announcement (2026); Shopify, AI commerce at scale (2026); Digiday, eCommerce ChatGPT-traffic conversion study (2026).