How do you get your products shown and cited in AI shopping answers?

AI shopping answers can only show products their crawlers can fetch and read. For ChatGPT, OAI-SearchBot has to reach your product pages and images, past both robots.txt and your CDN or firewall. Perplexity, Copilot and Google AI Mode depend on PerplexityBot, Bingbot and Googlebot. Each product page then needs its title, price, stock, brand and image in readable form. Blocking training bots like GPTBot doesn't remove you.

Nobody can promise a store a place in ChatGPT or AI Mode. You control whether these systems can reach your products, and how complete the product data is once they do.

Which AI crawlers need access to my product pages?

AI assistants use three kinds of bot, and they do different jobs. Search crawlers build the index that answers cite from. User-triggered fetchers read a page live when someone asks about it. Training crawlers collect text for model training. Only the first two affect whether shoppers see your products.

Crawler	Operator	What it's for	If you block it
OAI-SearchBot	OpenAI	ChatGPT search index and citations	Your site shows up in ChatGPT search only as a navigational link
PerplexityBot	Perplexity	Surfacing and linking sites in Perplexity results	Perplexity can't index your pages
Bingbot	Microsoft	Bing index; Copilot's results draw on the web	Bing can't index your pages
Googlebot	Google	Search, AI Overviews, AI Mode, the Shopping Graph	You lose Google Search and Google's AI features
ChatGPT-User, Perplexity-User, Claude-User	OpenAI, Perplexity, Anthropic	Live page reads when a person asks about a page	Assistants can't open a page a shopper points them to
GPTBot, ClaudeBot, CCBot, Google-Extended	OpenAI, Anthropic, Common Crawl, Google	Model training	No effect on search or shopping inclusion

As of 2026-10-17, OpenAI says OAI-SearchBot is used to surface websites in ChatGPT's search features, and that sites which block it appear only as navigational links. Perplexity describes PerplexityBot as designed to surface and link websites in its search results. Microsoft says Copilot's organic results draw on both the web and merchants' Microsoft Merchant Center feeds.

Should I block GPTBot?

That's your choice, and it doesn't decide shopping visibility. GPTBot collects training data. ChatGPT's shopping and search results depend on OAI-SearchBot. OpenAI also says ChatGPT-User, which fetches pages a user asks about, is not used to decide whether content may appear in search. smbuild reports a GPTBot block as information, not a failure.

The common mistake is a blanket "block all AI bots" list that catches OAI-SearchBot and PerplexityBot along with GPTBot. Name each crawler on purpose instead. A minimal robots.txt that allows the search tier and declines training:

User-agent: OAI-SearchBot
User-agent: PerplexityBot
User-agent: Claude-SearchBot
Disallow: /cart
Disallow: /checkout

User-agent: GPTBot
User-agent: ClaudeBot
User-agent: CCBot
User-agent: Google-Extended
Disallow: /

Sitemap: https://example.com/sitemap.xml

Under the robots.txt standard (RFC 9309), a crawler follows the most specific group that names it and ignores User-agent: *. So repeat any Disallow lines you still want, like cart and checkout paths, inside the named group.

Rules for anthropic-ai and Claude-Web name obsolete tokens and do nothing. If robots.txt returns a 5xx error or doesn't respond, crawlers treat the whole site as disallowed.

You can also declare preferences with a Content-Signal line, such as Content-Signal: search=yes, ai-input=yes, ai-train=no, plus a matching IETF AIPREF Content-Usage line. They state what you allow; they don't replace the crawler rules.

Why can't ChatGPT see my products when robots.txt allows it?

robots.txt is only half the gate. A CDN, firewall or bot-protection product can refuse a crawler that robots.txt allows. The crawler gets a 403, a 429 or a challenge page instead of your product. Product images often sit on a separate image CDN with its own rules, and OpenAI's merchant guidance asks stores to let OAI-SearchBot fetch image URLs without CDN 403s.

If Cloudflare's pay-per-crawl is on, AI crawlers get HTTP 402. Crawlers that don't pay skip those pages, so they can't be cited from them.

To fix an edge block:

1. Find which product, image and category URLs return an error to the crawler, in your CDN or WAF logs.
2. On Cloudflare, set the search crawlers (OAI-SearchBot, Claude-SearchBot, PerplexityBot) to Allow in AI Crawl Control, and exempt them from Bot Fight Mode and custom WAF rules.
3. On other WAFs, allow these verified bots by the operator's published IP ranges or Web Bot Auth signatures, not by the user-agent string alone. OpenAI publishes OAI-SearchBot's ranges at openai.com/searchbot.json.
4. Apply the same rules to the image CDN, not only the store domain.
5. If you use pay-per-crawl, decide per section whether crawl revenue or citation visibility matters more, and exempt search and user-fetch crawlers where visibility wins.
6. Let user-triggered fetchers (ChatGPT-User, Claude-User, Perplexity-User) through at the edge too, so assistants can read a page a shopper asks about.

What product data do AI shopping answers need from my pages?

Once a crawler gets in, it needs readable facts. OpenAI says ChatGPT shopping uses structured metadata from first- and third-party providers, such as price and description, plus reviews. Google says structured data isn't required for generative AI search, but recommends Merchant Center for visibility in AI responses.

OpenAI's product feed format, as of 2026-10-17, requires nine fields: item_id, title, description, url, brand, seller_name, image_url, availability and price. Checkout eligibility also needs privacy policy and terms of service links. Feed onboarding is open only to approved partners, but any feed or agent is built from these same facts.

To make that data derivable from the site:

1. Put Product markup on every product page, with a SKU or GTIN, name, description, brand, image, price and availability.
2. Make sure the markup matches the visible price and stock state.
3. Name the seller in Organization markup and in the home page title.
4. Link a privacy policy and terms of service from the footer, and state the return window in days on the returns page.

For Google's AI surfaces, Merchant Center has optional attributes used across AI-driven surfaces like AI Mode, including question_and_answer, document_link and related_product. They need raw material on the page: a product Q&A section, linked spec sheets or manuals (PDF), and related or accessory products.

ChatGPT also uses third-party content such as reviews. No source sets a threshold, so smbuild treats it as context, not a pass or fail.

Do I need a UCP profile, llms.txt or an agent-commerce protocol?

None of these is required to appear in AI shopping answers today.

UCP. Under the Universal Commerce Protocol, businesses publish a profile at /.well-known/ucp. Google's UCP integration for agentic checkout uses it, for select merchants in the US, Canada and Australia as of 2026-10-17. A valid profile is JSON over HTTPS with a top-level ucp object holding a version date, services and payment_handlers (required even if empty). Most stores don't have one yet, so absence is informational. Where your platform offers UCP support, enable it there.

llms.txt and agents.md. Google says you don't need AI text files and that Search doesn't use them (as of its 2026-07-10 update). No OpenAI source says ChatGPT shopping uses llms.txt. If you publish one as an agent convenience, follow the llmstxt.org format.

Checkout protocols. ChatGPT merchant approval and ACP, Microsoft Merchant Center for Copilot Checkout, and Google's agentic checkout all live in merchant accounts, not on your public site. As of 2026-10-17, Microsoft lists Copilot Checkout for US buyers paying in USD, and ChatGPT's Instant Checkout is a limited beta. Shopify says its merchants are included in ChatGPT shopping through Shopify Catalog (a vendor statement); check the Agentic Storefronts settings in your Shopify admin to see which channels are on.

How do I check if ChatGPT can see my products?

1. Open /robots.txt and find the group that applies to OAI-SearchBot. Confirm it doesn't disallow your product paths or image paths. Repeat for PerplexityBot, Bingbot and Googlebot.
2. Confirm robots.txt itself returns 200, not a 5xx or a timeout.
3. Check your CDN or WAF logs for OAI-SearchBot requests and the status codes they received. Logs from the real crawler, matched against OpenAI's published IP ranges, are the most reliable evidence.
4. Fetch one product page and one product image with the OAI-SearchBot user agent and look for a 403, 429, 402 or challenge page. A spoofed user agent from your own IP is a rough test: CDNs that verify bots by IP may block it while letting the real crawler through.
5. View the page source of a product page and confirm the price, availability, brand and image are there.

How smbuild checks this

smbuild reads your robots.txt, tests sampled product and image URLs against each crawler's rules, and fetches pages with the crawlers' user agents from smbuild's own IPs. Behind CDNs that verify bots by IP, edge blocks are reported with medium confidence. Product data is read from the server HTML of sampled product pages. Items that live in merchant accounts are marked Not measured, never scored as failures.

• AI-02 OAI-SearchBot can fetch products and images (ChatGPT shopping)
• AI-03 PerplexityBot, Bingbot and Googlebot allowed
• ORG-10 robots.txt lets search crawlers fetch products
• AI-01 Structured product data quality (shared prerequisite)
• AI-07 OpenAI product feed fields derivable from the site
• AI-06 Raw material for AI Mode conversational attributes
• AI-10 Third-party review and reputation footprint
• AI-04 UCP discovery profile (/.well-known/ucp)
• AI-09 llms.txt / agents.md (informational)
• AI-08 ChatGPT merchant approval / ACP, Microsoft Merchant Center (Copilot Checkout), Perplexity
• AI-05 Google agentic checkout eligibility (native_commerce, Google Pay & Wallet Console)

Sources

• OpenAI: Overview of OpenAI crawlers (checked 2026-10-17)
• OpenAI: Commerce product feed spec (checked 2026-10-17)
• OpenAI: Commerce key concepts (checked 2026-10-17)
• OpenAI: Commerce get started (checked 2026-10-17)
• OpenAI Help: Shopping with ChatGPT search (checked 2026-10-17)
• OpenAI: Search product discovery (checked 2026-10-17)
• Perplexity: Perplexity crawlers (checked 2026-10-17)
• Microsoft Advertising: Agentic commerce (checked 2026-10-17)
• Google Search Central: AI optimization guide (updated 2026-07-10)
• Google Merchant Center: Conversational attributes for AI surfaces (checked 2026-10-17)
• Google Merchant Center: Agentic checkout (checked 2026-10-17)
• Google Merchant: UCP integration guide (updated 2026-10-15)
• Universal Commerce Protocol: Specification overview (checked 2026-10-17)
• Shopify: Agents (checked 2026-10-17)
• RFC 9309: Robots Exclusion Protocol
• Cloudflare: Pay per crawl
• Content Signals
• llms.txt specification

Questions

Does blocking GPTBot remove my products from ChatGPT shopping?

No. GPTBot collects training data. ChatGPT search and shopping depend on OAI-SearchBot, so you can block GPTBot and still allow OAI-SearchBot.

What happens if I block OAI-SearchBot?

OpenAI says sites that block OAI-SearchBot appear in ChatGPT search only as navigational links. Your product pages and images can't be fetched for shopping results.

My robots.txt allows AI crawlers. Why can't they reach my store?

A CDN, firewall or bot-protection rule can still refuse them with a 403, 429, 402 or a challenge page. Check your CDN logs and allow the verified search crawlers on both the store domain and the image CDN.

Do I need llms.txt to show up in AI shopping answers?

No. Google says Search doesn't use AI text files, and no OpenAI source says ChatGPT shopping uses llms.txt. It's optional.

What is a UCP profile and do I need one?

It's a JSON file at /.well-known/ucp that describes your store's Universal Commerce Protocol services. Google's agentic checkout for select US, Canada and Australia merchants uses it, but most stores don't have one and it isn't needed to appear in AI answers.

Can anyone guarantee my products appear in ChatGPT or AI Mode?

No. You can make sure the crawlers can reach your products and that each page carries complete product data. Whether a product is shown for a given question is decided by each platform.