How to get your store recommended by ChatGPT: a guide for Shopify and WooCommerce

More shopping journeys now start with a sentence in a chat window: "find me waterproof trail shoes under $100". The assistant answers with a few products and a few stores. The question for you is simple. Is your store one of them?

This guide explains how assistants pick stores, what to fix first, and how much traffic is really at stake.

How big is this, really?

Start with honest numbers. AI referred traffic is still a small slice: Similarweb put it at 0.89% of ecommerce traffic in May 2026. It is growing fast though. Adobe measured a 393% year over year jump in AI traffic to US retail sites in Q1 2026, and those visitors converted better than average.

So this is not where most of your revenue comes from today. It is a channel that is cheap to win now and expensive to win later, when every competitor has done the basics.

How an assistant picks a store

An AI assistant works with data, not with impressions. Before it recommends you, three things must happen:

• It gets in. Its crawler or live fetcher can load your pages. robots.txt, your CDN and your firewall decide this.
• It understands. It can read what you sell, the price, the currency and whether it is in stock. Structured data and feeds decide this.
• It trusts. It can see who you are, how to contact you, and your shipping and returns terms.

Fail any of the three and you do not rank lower. You simply do not show up.

What is live in the US right now

Agentic checkout is real in the US, but narrower than the headlines suggest. OpenAI launched Instant Checkout in 2025 and reworked it in March 2026: ChatGPT now finds products through merchant feeds and sends the buyer to your site. Shopify opened checkout to browser agents in September 2026. Microsoft runs Copilot Checkout in the US.

In the UK and most other markets, assistants recommend and link, but do not buy for the shopper yet. Either way, the first step is the same: get read and get recommended.

The four fixes with the most impact

1. Let AI bots in

Check that robots.txt does not block GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot or PerplexityBot. Then check the layer people forget: your CDN. Cloudflare changed its defaults on September 15, 2026, and some stores now block AI agents without knowing it. We wrote a full post on how to check if Cloudflare is blocking ChatGPT.

Time: 15 to 30 minutes.

2. Publish product structured data

Every product page should carry a schema.org Product with an Offer: name, image, price, priceCurrency and availability.

• Shopify: most current themes include it. Check it survived your theme edits.
• WooCommerce: core outputs it. Yoast WooCommerce SEO or Rank Math add brand and GTIN. Make sure only one plugin outputs the Product block.

Our post on product schema for AI shopping agents has a copy and paste example.

Time: one or two evenings if your theme is missing it.

3. Clean prices and stock

Price as a plain number, currency as an ISO code, availability that matches reality. A price in the markup that differs from the price on the page is a trust problem for Google and for any agent that compares them.

Time: about an hour.

4. Send a feed

Connect Google Merchant Center. It is free, and it is the official door to Google Shopping. On Shopify, use the Google and YouTube app. On WooCommerce, use Google for WooCommerce. In the US, OpenAI and Perplexity also accept merchant feeds.

Time: an afternoon for the first setup.

Nice to have: llms.txt and agents.md

Two small text files at the root of your site:

• llms.txt lists your key pages with a short description. Google has said it ignores the file, so do not expect a ranking boost. It is cheap, and some AI agents read it.
• agents.md tells AI agents what they can do on your store and under which rules.

Both take minutes with our free generators: llms.txt and agents.md. Shopify already serves a native llms.txt since May 2026.

Trust signals assistants look for

Shoppers ask assistants things like "is this store legit?" and "do they ship to the UK?". Make the answers easy to find:

• A clear About page with a real business name and address.
• Contact details on every page, usually in the footer.
• Shipping and returns pages written in plain language, with numbers: days, costs, countries.
• An Organization block in your structured data, with name, logo and social profiles.

Mistakes we see every week

We scan stores every day. These problems come up again and again:

• A firewall that blocks bots the owner never knew about. The store loads fine in a browser, but GPTBot or ClaudeBot gets a 403.
• Prices that only exist in the image or in JavaScript. If the price is not in the HTML or the structured data, an assistant may not see it at all.
• Out of stock products marked InStock. An assistant that sends a shopper to a dead end learns not to trust you.
• Thin product copy. "Premium quality, great value" tells a machine nothing. Size, material, fit and use case tell it a lot.
• Missing contact details. No address, no email, no phone. Assistants are cautious about stores they cannot verify.

None of these need a developer. Most take less than an hour to fix once you know they exist.

Check where you stand today

Paste your URL into the free AISaysYes scanner. You get a 0 to 100 score in about 30 seconds, with every check explained and the time each fix takes. No sign-up.

Then use the step by step guide to work through the fixes by impact. If you want to see how stores in your niche score, look at the public ranking.

Stores that already passed the scan