AI search optimization for online stores: how to get recommended by ChatGPT, Gemini and Perplexity, and what is hype

Shoppers now ask ChatGPT, Gemini, Perplexity and Google's AI Mode which product to buy, and store owners want to know how to be in the answer. A whole industry now sells AI search optimization, also called generative engine optimisation or GEO. Some of the advice is sound, some is ordinary SEO with a new label, and some is guesswork sold with confidence.

This guide is for owners of Shopify, WooCommerce and custom stores who want to separate what is documented from what is hype: where each assistant gets product information, the levers you control, where llms.txt fits, and how to measure AI referral traffic. Platform claims are current as of 15 October 2026.

I build e-commerce automation and AI agents. I do not sell rankings, and nobody can honestly promise a place in an assistant's answer. For AI checkout and the protocols behind it, read my guide to agentic commerce for online stores; this post is about being found.

Where ChatGPT, Gemini, Perplexity and AI Mode get product information

Each assistant combines a web index with structured product data. The mix differs.

Assistant	Main product sources	What you control
ChatGPT	Its own search crawler, merchant product feeds, Shopify's catalogue, third-party product data	Crawler access, feed (if accepted), page content
Google AI Mode and Gemini	Google's search index and Shopping Graph, fed by Merchant Center	Merchant Center feed, structured data, page content
Perplexity	Its own crawler, a merchant programme, Shopify integration (US)	Crawler access, page content
Microsoft Copilot	Bing's index; Copilot Checkout merchants (US)	Bing indexing, page content

A few details matter:

• ChatGPT. OpenAI's help centre says product results are selected independently by ChatGPT and are not ads, and that it draws on structured metadata from merchants and third-party providers, such as price and description. OpenAI's crawler documentation says sites that opt out of OAI-SearchBot "will not be shown in ChatGPT search answers", and that robots.txt changes take about 24 hours to apply.
• Google. In May 2025 Google said its Shopping Graph held more than 50 billion product listings, with more than 2 billion refreshed every hour. Merchant Center feeds are how your products get into it.
• Perplexity. Its documentation describes PerplexityBot, which surfaces sites in results, and Perplexity-User, which fetches pages a user asked about. Its merchant programme, launched in November 2024, is free but was aimed at large retailers.

Assistants read what is already published about your products: your pages, your feeds, and what other sites say about you.

AI search optimization: the levers you control

None of these are secret. They are the same things that make a store easy for search engines and pleasant for people.

1. Let the search crawlers in, separately from training crawlers

Search bots and training bots are different user agents. You can allow OAI-SearchBot and PerplexityBot so you appear in answers, and block GPTBot if you do not want your pages used for training. Google says ordinary Googlebot rules control AI Overviews and AI Mode, and that Google-Extended governs training and grounding in its other AI systems. Check your CDN's bot settings too, since some block AI crawlers before robots.txt is read.

2. Keep product data complete in your feeds

Titles that say what the item is, every attribute filled, GTINs where the manufacturer assigned them, and price and stock that match the page. Assistants compare products on facts; a missing fact leaves you out. My guide to product feed management covers the fields Google, Meta, TikTok and OpenAI require.

3. Mark up pages, and keep the markup honest

Google's guidance on AI features says there is "no special schema.org structured data" needed to appear in AI Overviews or AI Mode, and it recommends making sure your structured data matches the visible text. In practice, that means Product markup with price, availability and reviews, plus organisation-wide return and shipping policies, which Google added to its markup support in 2024 and 2025 and which you can also set in Search Console.

4. Earn reviews and mentions you can stand behind

Assistants summarise what others say. Collect real reviews on your product pages and through Merchant Center's product reviews feed, and answer questions where customers ask them. Fake reviews break platform rules and lose trust fast when an assistant quotes them back.

5. Write facts an assistant can quote

Delivery times per country, returns window, who pays return postage, warranty, sizing, compatibility, materials. One clear sentence beats a paragraph of brand voice. If a shopper asks "does this fit a 13-inch laptop", the answer should be written down on the page.

What is hype

• Guaranteed placement. No one controls what ChatGPT or Gemini recommends. OpenAI says its product results are not ads, so there is nothing to buy there.
• Special AI files for Google. Google says you do not need to create new machine-readable files, AI text files or markup to appear in its AI features, and that there are "no additional requirements" beyond normal search eligibility.
• Rewriting every page "for AI". Filling in missing facts helps. Rewriting good copy into stilted question-and-answer blocks rarely does.
• Prompt text hidden in pages. Instructions aimed at language models and hidden from people are a trick with an obvious risk: Google's spam policies already cover hidden text.
• Dashboards that "rank" you in ChatGPT. Tools that run sample prompts can show trends, but answers vary by user and conversation. Treat their scores as a rough signal.

llms.txt: a proposal, not a standard

llms.txt is a proposal published by Jeremy Howard in September 2024: a Markdown file at the root of a site that gives language models a short guide to its content, with links to clean Markdown versions of key pages. It is easy to add and harmless. It is also unproven. Google's John Mueller wrote in June 2025 that no AI system currently used llms.txt, as reported at the time, and Google's AI features guidance says no AI text files are required. Adoption by the other assistants is unclear, and none of them has documented using it for ranking.

My own site is a small example of the cheap end of this. devwren.com publishes an llms.txt file, returns any page as Markdown when a client asks for it, carries structured data, and runs a read-only MCP server for AI agents, all listed on the developers page. I did it because it costs little and helps an agent that is already on the site read it accurately. I would not tell a store owner to expect more traffic from it.

How to measure AI referral traffic

Measure before you spend. Adobe data reported by TechCrunch showed AI traffic to US retail sites up 393% year on year in Q1 2026, from a small base, so expect a small but growing channel.

• Analytics referrals. In GA4 or similar, build a channel group for referrers such as chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. OpenAI's publisher FAQ says ChatGPT adds utm_source=chatgpt.com to the links it sends.
• Search Console. Google counts AI Overviews and AI Mode inside the Performance report's Web search type rather than as a separate report.
• Bing Webmaster Tools. Microsoft launched an AI Performance report in public preview on 10 February 2026, showing how often your pages are cited in Copilot and Bing's AI answers.
• Server logs. Count requests from OAI-SearchBot, PerplexityBot and ChatGPT-User. If the search bots never visit, nothing else on this list matters yet.
• Revenue, not just visits. Tag orders by referral channel before deciding how much effort AI search deserves.

Some app traffic arrives without a referrer and looks like direct traffic, so treat these numbers as a floor.

A 30-day plan for a small store

1. Week 1: access. Check robots.txt and CDN bot settings for the search crawlers. Confirm your pages render product facts without needing scripts to run.
2. Week 2: data. Fix Merchant Center errors, fill missing attributes and make sure price and stock match the page.
3. Week 3: facts and markup. Rewrite shipping, returns and sizing pages in plain sentences, and validate your Product and policy markup.
4. Week 4: measurement. Set up the AI referral channel group, check Search Console and Bing, and note a baseline to compare against next quarter.

What AI search optimization costs: platform, or automation

• On Shopify, much of this is handled: Agentic Storefronts, active by default for eligible stores, send your catalogue to ChatGPT, Copilot and Google's AI surfaces. Your work is the data and the policies.
• On a custom store or several channels, you need a feed generated from one catalogue, structured data that matches it, and stock that stays accurate. That is automation work. My marketplace and e-commerce plans on the pricing page start from $2,500 for an MVP with one automation, such as listing updates; these are starting prices, and every project gets a fixed quote after a short brief. The e-commerce automation page shows the kind of work involved.
• An assistant on your own site that answers product questions from your catalogue is custom AI agent work, quoted from a brief.

FAQ

Does llms.txt help my store appear in ChatGPT or Google AI Mode?

There is no evidence that it does. llms.txt is a proposal from September 2024, not a standard any assistant has documented using for ranking. Google says its AI features need no special AI text files, and Google's John Mueller wrote in June 2025 that no AI system currently used the file. Adding one is cheap and harmless, but crawler access, complete product data and clear policies matter far more.

Can anyone guarantee my products will be recommended by ChatGPT or Gemini?

No. OpenAI says ChatGPT's product results are selected independently and are not ads, and Google says there are no special optimisations for its AI features beyond normal search eligibility. Answers also change with each shopper's question, location and history. What you can control is whether assistants can crawl your pages, whether your product data is complete and accurate, and whether your policies and reviews are clear enough to quote.

How do I see how much traffic AI assistants send my store?

Create an analytics channel group for referrers such as chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com, and look for utm_source=chatgpt.com, which OpenAI says ChatGPT adds to its links. Google counts AI Mode and AI Overviews inside Search Console's Web search type, and Bing Webmaster Tools has an AI Performance report for Copilot citations. Some app visits arrive without a referrer, so treat the totals as a minimum.

If you want help making a store easier for assistants to read, send me a short brief: your platform, which channels you sell on, roughly how many products, and where your product data lives. You can send it through the contact page, read about the AI agents I build for businesses, or start with my guide to e-commerce automation if you sell on several channels.