A growing share of shopping starts with a question typed into an AI assistant: “best golf glove for wet weather,” “running shoes for flat feet under $150,” “a gift for someone who likes hiking.” The assistant answers with a short list of products and links. If your store is not on that list, you never get the click.
Nobody outside the AI companies knows exactly how each assistant ranks products, and it changes often. But the pattern across ChatGPT, Perplexity, Gemini, and Google’s AI Overviews is consistent enough to act on. They recommend what they can read, what they can trust, and what answers the question that was asked.
1. They can only recommend what they can read
Each assistant fetches pages with its own crawler. If your robots.txt blocks those crawlers, or an app blocks them on your behalf, you are invisible to that assistant no matter how good your products are. The ones that matter most for shopping answers are the search crawlers: OAI-SearchBot and ChatGPT-User for ChatGPT, and PerplexityBot for Perplexity. Blocking the training-only crawlers is a separate choice and does not keep you out of search answers.
The second part of “readable” is server-rendered text. Many crawlers do not run JavaScript. If your product descriptions, reviews, or size charts only appear after scripts load, the crawler sees an empty page.
2. They trust structured facts
Price, availability, brand, ratings, and identifiers like GTIN or SKU are much easier for an assistant to use when they are published as structured data (Product schema in JSON-LD) rather than buried in page copy. Shopify themes include some of this by default, but it is often incomplete: missing availability, no brand, or no review rating because the reviews app does not output it.
Ratings matter more than most brands think. An assistant asked for “the best” anything leans toward products with visible, verifiable reviews.
3. They quote pages that answer the exact question
A product page that says “premium quality, great fit” gives an assistant nothing to quote. A page that says who the product is for, what it is made of, how it fits compared with alternatives, and how to care for it gives the assistant the specifics it needs to match your product to a question.
Buying guides and comparison pages work the same way. “How to choose a golf glove for hot weather” is a question people actually ask, and a page that answers it well, with your products as examples, is a page assistants can cite.
4. They lean on what others say about you
Assistants weigh reviews, comparison articles, forums, and press. You cannot control those directly, but you can make it easy to be included: clear brand information, consistent product names, and an Organization schema that connects your social profiles to your store.
Where to start
- Check robots.txt and make sure the AI search crawlers are allowed.
- Complete your Product schema, including availability and review ratings.
- Rewrite your top ten product pages with real specifics.
- Write three buying guides around the questions your customers ask most.
- Pick twenty buyer questions and test them in ChatGPT and Perplexity every month, so you know whether anything is changing.
Our free AI Readiness Check runs the technical part of this list against your store in about 30 seconds. If you want the whole program handled, including content and monthly tracking, see AI Search Visibility.
Scott founded Broke Hold. He works hands-on across ecommerce operations, automation, cloud infrastructure, and integrations, works as the fractional CTO for Shopify brands, and builds and runs Broke Hold's own software, including VPS Snaps and DingPack. He served in the U.S. Navy Reserve.