An AI visibility scan reveals five things about your Shopify store: which AI crawlers can actually reach it, whether your product data is machine-readable, whether an AI shopping agent could transact with it, whether your pages render without JavaScript, and whether emerging files like llms.txt exist. Those five categories combine into one 0–100 AI-readiness score.
Most merchants have never seen any of these numbers, and the gap is expensive. Otterly’s AI Citations Report 2026, published February 1, 2026 after analyzing more than one million AI citations across ChatGPT, Perplexity, and Google AI Overviews, found that 73% of sites had technical barriers blocking AI crawler access. Three quarters of the web is failing before content quality is even considered.
On this page
- What is an AI visibility scan?
- What does an AI visibility scan check?
- 1. Which AI crawlers can actually reach your store
- 2. Whether your product data is machine-readable
- 3. Whether an AI shopping agent could buy from you
- 4. Whether your pages are readable without JavaScript
- 5. Whether your emerging AI files exist — and whether they matter
- How to fix what the scan finds
- Frequently Asked Questions
What is an AI visibility scan?
An AI visibility scan is a diagnostic that inspects the machine-readable signals AI assistants use to find, understand, and recommend your store, then scores each one. It is not an SEO audit. An SEO audit asks whether Googlebot can rank your pages; an AI visibility scan asks whether ChatGPT, Claude, Perplexity, and Gemini can read your store at all — and whether what they read is complete enough to quote.
The distinction matters because the two systems consume different inputs. A Shopify store can rank on page one of Google and still be invisible in AI answers, because AI assistants depend on crawler permissions, structured data, and server-rendered HTML rather than backlinks and keyword relevance. The scan surfaces the failures that a rankings report will never show you.
Otterly’s report puts the priority plainly: “A perfectly written article that AI cannot access helps nobody. An adequately written article that AI can easily parse and cite will outperform your masterpiece every time.”
What does an AI visibility scan check?
The AI Visibility Checker grades five categories and weights them by how strong the real-world evidence is that each one affects visibility. A Shopify storefront is scored like this:
| # | Category | Weight | Evidence strength | What it reveals |
|---|---|---|---|---|
| 1 | AI Crawler Access | 30% | Strong | Whether ChatGPT Search, Claude, Perplexity, Google, Bing, and Amazon crawlers can fetch your pages — in robots.txt and at your CDN |
| 2 | Structured Data | 30% | Strong | Whether your Product, Offer, and Organization JSON-LD carry the fields AI needs to quote you |
| 3 | Commerce / Agentic Readiness | 20% | Medium | Whether an AI shopping agent can find a product page and read price, availability, and image |
| 4 | Crawlability & Content | 15% | Medium | HTTPS, sitemap, canonical, heading structure, and whether real text exists in the server HTML |
| 5 | Emerging AI Files | 5% | Weak | Whether llms.txt and agents.md are published — deliberately weighted low |
Two things about that table are worth stating outright. First, the weights are not evenly split, because the underlying signals are not equally proven. Second, the fifth category is intentionally worth only 5% — most tools in this category inflate llms.txt into a headline metric it has not earned.
Want your store’s five numbers? Run the free AI Visibility Checker — no login, no install, a full category breakdown in about 60 seconds.
1. Which AI crawlers can actually reach your store
The first thing a scan reveals is whether AI search crawlers are allowed to fetch your store — and it checks two separate layers, because a store can pass one and fail the other. The scan reads your robots.txt rules per user-agent, then sends a live request as each crawler to see what your CDN or firewall actually returns.
That second layer catches the failure merchants never find manually. If robots.txt says Allow but your WAF answers a PerplexityBot request with HTTP 403 — or 401, 405, 429, 451, or 503 — you are blocked in practice while looking perfectly configured on paper. Bot-mitigation rules, aggressive rate limits, and security apps all cause this silently.
The scan also separates crawlers that affect visibility from crawlers that only affect training:
- Affects whether you appear in answers:
OAI-SearchBot(ChatGPT Search),ChatGPT-User(in-chat fetches),Claude-SearchBot,PerplexityBot,Googlebot,Bingbot,Amazonbot. - Affects model training only:
GPTBot,ClaudeBot,Google-Extended,meta-externalagent,CCBot.
This distinction is the single most common misconfiguration in Shopify robots.txt files. Merchants who read a 2024 blog post about “blocking AI scrapers” often disallowed GPTBot — a training crawler — and assumed they were done, while others copied a blanket Disallow that also caught OAI-SearchBot and quietly removed their store from ChatGPT Search results. Blocking training bots is a legitimate business choice; blocking search bots costs you customers.
2. Whether your product data is machine-readable
The second category grades your structured data — the JSON-LD that tells an AI what you sell in a format it does not have to guess at. The scan parses the JSON-LD on your homepage and a product page, then checks for Product schema and five commercial fields inside it: price, price currency, availability, brand, and GTIN/MPN. It also credits Organization, BreadcrumbList, AggregateRating, and FAQPage markup.
Default Shopify themes usually emit some Product schema, which is why merchants assume this category is safe. It rarely scores full marks. The typical failure is partial: a Product node with a name and image, no priceCurrency, no availability, and no gtin. An assistant asked “is this in stock and how much?” cannot answer from that, so it recommends a competitor whose markup does answer.
Be aware of an honest caveat here. Whether schema directly causes AI citations is genuinely contested. Ahrefs tracked 1,885 pages that added JSON-LD schema between August 2025 and March 2026, matched them against 4,000 control pages, and found citations moved 2.4% on Google AI Mode and 2.2% on ChatGPT — no meaningful lift. The reason this category still carries 30% weight is different and more concrete: agentic shopping surfaces need machine-readable price and availability to quote or transact at all. Complete schema is table stakes for commerce, not a citation trick.
3. Whether an AI shopping agent could buy from you
The third category simulates the path an AI shopping agent takes. The scan starts at your homepage, follows links to find a crawlable product page, and then checks whether that page exposes what a purchase decision requires: a price an agent can read, an availability state, a product image in structured data, a sitemap.xml for discovery, and a canonical URL so duplicate variant URLs do not confuse the agent.
A store fails this category most often at step one — no product page is reachable from the homepage without JavaScript, so the agent never gets to the product data at all.
This is the category with the shortest history and the most movement — which is precisely why it rewards fundamentals over any single integration. OpenAI and Stripe’s Agentic Commerce Protocol gave agents a standard way to complete purchases, but OpenAI pulled back from native in-chat checkout in early 2026. As an OpenAI spokesperson told Modern Retail on March 12, 2026: “We’re prioritizing making ChatGPT search and product discovery great, with the Agentic Commerce Protocol serving as the infrastructure that connects users to merchants across the full shopping journey.” In practice, buyers now discover in the chat and complete the purchase on the merchant’s own storefront — which makes your on-site product data more load-bearing, not less. It is also exactly what our guide to getting your store ready for AI shopping agents walks through in depth.
4. Whether your pages are readable without JavaScript
The fourth category checks crawl fundamentals: HTTPS, a reachable sitemap.xml, a canonical tag, exactly one H1, supporting H2 headings, visible FAQ-style content, and — the one that surprises people — whether your server-rendered HTML contains real text at all.
Most AI crawlers do not execute JavaScript. If your homepage ships an empty shell and paints its content client-side, the crawler receives a nearly blank page. The scan flags a homepage that returns under roughly 600 characters of readable server HTML, because that is the practical threshold below which an assistant has nothing to summarize.
Industry data confirms how widespread this is. Adobe Analytics data reported by Digital Commerce 360 on June 17, 2026 scored retail pages for AI readability by category and found cosmetics at 63%, electronics at 56%, sporting goods and apparel at 51% each, grocery at 48%, and furniture and home at 47%. Roughly half the content on a typical retail site cannot be read reliably by an AI model. The same Adobe data found AI-referred traffic converted 54% better than non-AI sources in May 2026 — so the readable half of the web is capturing a disproportionately valuable audience.
5. Whether your emerging AI files exist — and whether they matter
The fifth category checks for llms.txt and agents.md, and it is capped at 5% of your score on purpose. This is the honest part of the report: publishing llms.txt is cheap and harmless, but the evidence that it drives AI visibility today is thin.
The data is unambiguous. Ahrefs analyzed 137,210 domains in May 2026 and found that while 28% publish an llms.txt file, 97% of those files received zero requests during the month. The study’s conclusion: “If you publish an llms.txt file today, the most likely outcome by far is that nothing ever fetches it.” Google’s John Mueller was equally direct in comments reported June 2, 2026: “I don’t think anyone knows – it’s purely speculative for now (the file has existed for years, yet none of the AI systems use it — what does it mean?).”
So why grade it at all? Because the cost of publishing one is close to zero, the file becomes useful the moment an assistant does start reading it, and a stale or missing file is a reliable proxy for how well a store maintains its other AI signals. If you want the full picture on the format, our guide to what llms.txt is for Shopify merchants covers the spec and the realistic upside. A scan that presents llms.txt as 30% of your AI readiness is selling you something; 5% is what the evidence supports.
Your five category scores roll up into a single 0–100 number with four bands: Not AI-ready below 50, Needs work at 50–74, Good at 75–89, and Excellent at 90+. For how those bands compare across stores, see our breakdown of what counts as a good AI visibility score in 2026.
How to fix what the scan finds
A scan is a diagnosis, not a cure — and the fixes it prescribes decay. Theme updates rewrite robots.txt. A new security app adds a WAF rule that starts returning 403 to PerplexityBot. Products ship without gtin. Your llms.txt goes stale the day you add a collection. Merchants who fix everything by hand in an afternoon routinely score lower three months later than they did the week after the cleanup.
Kedra AI Index exists to hold those five categories green without a monthly manual audit. Concretely, it maps to the scan category by category:
- AI Crawler Access — allow or block individual crawlers (
OAI-SearchBot,ClaudeBot,PerplexityBot, and the rest) with per-bot toggles, so you can permit search crawlers while restricting training crawlers. - Structured Data — reinforces your Product and Offer markup so price, availability, and brand stay exposed as your catalog changes.
- Commerce / Agentic Readiness — keeps product data complete and discoverable for AI shopping agents.
- Crawlability & Content — runs an AI Readiness Check with a plain checklist of what is working and what needs fixing.
- Emerging AI Files — generates and hosts a live
llms.txton your domain and auto-refreshes it as products and prices change.
It also answers the question no free scan can: which assistant actually referred each session and order, so you can see AI revenue rather than guessing at it.
Ready to fix what your scan flagged? Install Kedra AI Index from the Shopify App Store — it’s free to start.
Frequently Asked Questions
How long does an AI visibility scan take?
About 60 seconds. The AI Visibility Checker fetches your homepage, robots.txt, sitemap.xml, and a product page, probes each major AI crawler live, parses your JSON-LD, and returns a 0–100 score with a category-by-category breakdown. No login, no app install, and nothing is added to your theme.
Is an AI visibility scan the same as an SEO audit?
No. An SEO audit measures ranking factors like keywords, backlinks, and Core Web Vitals for Googlebot. An AI visibility scan measures whether AI assistants can access and parse your store: crawler permissions at both robots.txt and CDN level, structured data completeness, server-rendered content, and commerce readiness. A store can score well on one and poorly on the other.
What is the most common problem an AI visibility scan finds?
Crawler blocking at the edge. A store’s robots.txt permits AI crawlers, but its CDN, firewall, or bot-mitigation app returns a 403 or 429 to the same request. Because robots.txt looks correct, merchants never suspect it. A live probe as each crawler is the only reliable way to catch this failure.
Does blocking GPTBot hurt my AI visibility?
Not directly. GPTBot is OpenAI’s training crawler; ChatGPT Search uses OAI-SearchBot and in-chat browsing uses ChatGPT-User. Blocking GPTBot keeps your content out of training data without removing your store from ChatGPT answers. Blocking OAI-SearchBot does remove you. A good scan reports these separately instead of lumping them together.
How often should I re-run an AI visibility scan?
Re-scan after every theme update, app install, or major catalog change, since each can silently reset robots.txt, strip structured data, or introduce an edge rule that blocks crawlers. Monthly is a reasonable floor for a stable store. Continuous monitoring is more practical than periodic manual scans, because these signals decay quietly between checks.
Your five numbers, then your next fix
An AI visibility scan converts a vague worry — “are we showing up in ChatGPT?” — into five specific, fixable numbers: crawler access, structured data, commerce readiness, crawlability, and emerging files. With 73% of sites carrying technical barriers to AI crawlers and roughly half of retail content unreadable to AI models, the merchants who measure first are the ones who get recommended.
Start with the diagnosis: scan your store’s AI visibility for free and see all five category scores in about a minute. Then, when you want those categories to stay green through every theme update and catalog change, install Kedra AI Index to maintain crawler access, reinforce your product schema, host a live llms.txt, and track how often ChatGPT, Claude, Gemini, and Perplexity send you actual revenue.
Kedra Team
Expert insights on Shopify development and e-commerce growth strategies.