Improve your store’s AI search readiness by checking crawler access, readable product information, accurate structured data, and consistent business details. Then measure actual mentions and citations separately. A technical audit can identify barriers to discovery, but neither a high score nor valid schema guarantees an AI recommendation.
A store can pass a technical audit and still be absent from a shopping answer. Readability makes you eligible to be considered; useful product information and relevant evidence give shoppers a reason to choose you.
Shoppers are changing where they start. More people now ask ChatGPT, Perplexity or Google’s AI answers “what is the best store for X” before they ever open a search results page. If your store is not part of that answer, you are not losing a ranking. You are missing from the conversation entirely.
Over the last three weeks we ran a free AI visibility check on hundreds of websites, and the results show that most businesses are not ready for this shift.
Between September 15 and October 3, 2026, 731 different websites ran our check. Most of them were small and mid-size businesses: online stores, agencies, clinics, service companies. Each site got a score from 0 to 100 based on what AI assistants actually need: can their crawlers read the pages, is
there enough clear text to quote, is the structured data in place, and does the site explain who the business is and what it sells.
The results:
1. AI crawlers are blocked by accident. Many robots.txt files, security plugins and CDN settings block bots like OAI-SearchBot, GPTBot or PerplexityBot. The store owner usually has no idea. If the assistant cannot read your pages, it cannot recommend you.
2. The text is not really there. Product pages built with heavy JavaScript can look perfect to a human and almost empty to a crawler. AI search bots generally do not run scripts the way a browser does, so they see a page with a title and little else.
3. Thin product and category pages. A product name, a price and three bullet points give an assistant nothing to quote. Pages that answer real buyer questions (sizing, materials, shipping times, returns, who the product is for) are the ones that get cited.
4. No structured data. Product, Organization and FAQ markup tells machines exactly what you sell, at what price, with which reviews. Without it, the assistant has to guess, and it usually picks a competitor that made it easy.
5. Nobody else talks about you. AI assistants lean heavily on what other sites say: reviews, comparisons, roundups, press mentions. A store that exists only on its own domain looks less trustworthy than one that is mentioned elsewhere.
You can test the basics without any tools:
Start with access, because nothing else matters if bots cannot read the site. Then add plain text answers to the questions buyers actually ask, directly on your top product and category pages. Add Product and FAQ structured data next. Finally, work on mentions outside your site: ask happy
customers for reviews on independent platforms, pitch your products to niche roundups, and answer questions in communities where your buyers spend time.
None of this replaces classic SEO. It sits on top of it. Stores that already have clean, fast, well-described pages are usually one or two fixes away from being readable by AI assistants.
Nine out of ten sites in our data are not fully prepared for AI search. That means the stores that fix the basics now will have an advantage while competitors are still guessing why their traffic looks different. Check the access, check the text, check the markup, then ask the assistant yourself.
The answer will tell you exactly where you stand.
Maksim Shchegolev is the founder of OperStack, a service that checks how AI assistants see and recommend websites.
AI search readiness depends on access and accurate information, while recommendation performance requires separate testing across relevant shopper questions.
Check whether AI search systems can read your store by reviewing crawler permissions, security responses, and the content delivered on important product pages. Start with robots.txt, then ask your technical owner to check CDN or firewall restrictions and actual crawler requests. Compare the initial HTML with the visible page to identify missing product facts. Disabling JavaScript is a useful diagnostic, but it does not reproduce every crawler. Finally, inspect indexing and rendered content in Google Search Console for Google’s search systems.
You do not need to allow GPTBot solely to participate in ChatGPT search, because GPTBot training preferences and OAI-SearchBot search access are separate controls. OpenAI identifies OAI-SearchBot as the relevant crawler for content appearing in search summaries and snippets. Review both settings according to your business policy rather than unblocking every AI-related agent. Also confirm that your security infrastructure permits legitimate search crawler requests. Allowing access improves discoverability, but it does not guarantee that ChatGPT will recommend or cite your store.
Blocking Google-Extended does not by itself remove your store from Google’s AI Overviews or AI Mode, because those features use Google Search eligibility and controls. Googlebot access, indexing, and eligibility to display a snippet are the relevant foundations. Google-Extended concerns separate training and grounding uses in some other Google systems. Treat those preferences independently. Before changing robots.txt, establish which outcome you intend to control, and avoid blocking Googlebot accidentally while attempting to manage a separate AI-related use.
Product or FAQ schema does not guarantee AI recommendations, and Google requires no special schema for AI Overviews or AI Mode. Product markup can help communicate facts such as price and availability, provided it matches the visible page. FAQ markup can describe genuine visible questions and answers, but it is not a required AI discovery mechanism. Validate the information you publish and correct inconsistencies across pages and feeds. Treat schema as part of information quality, not as a replacement for useful content, accessibility, or independent evidence.
Measure AI visibility improvement by repeating a consistent set of shopper questions and recording brand mentions, product appearances, citations, and factual accuracy separately. Start with 10 to 20 relevant prompts that include your target market and common purchase constraints. Record the platform and date, and compare repeated observations rather than relying on one answer. Add referral traffic and conversions where your analytics can identify them. Keep readiness scores separate: a better technical score shows audit improvement, not proof that recommendation frequency or revenue increased.
