The Front Door Test: Can AI Crawlers Even Get Into Your Store?

The Front Door Test: Can AI Crawlers Even Get Into Your Store?

Before an AI engine can recommend your store, it has to get inside it, and getting inside is a sequence of gates most merchants have never checked. You can have the best products on the internet and still be invisible, not because you lost the comparison, but because the crawler never reached it.

By Margareta Petrovic, founder of Visibility Mesh. We measure how legible ecommerce stores are to AI, and publish what we find. Updated June 2026.
Key takeaways
  • Roughly three in four businesses are estimated to be effectively invisible in AI search. Often an access problem, not a content one.
  • Most major AI crawlers don't run JavaScript. If your products load by script, they can read as a blank page.
  • Amazon blocks the ChatGPT crawlers in robots.txt; a clean Shopify store can appear in ChatGPT shopping where Amazon can't.
  • Four gates sit between a crawler and your products. A 'no' at any one ends the visit. Silently.

Everyone wants to talk about being recommended by AI. Almost nobody checks step one: can the machine get through the front door at all? By one widely cited 2026 estimate, roughly three in four businesses are effectively invisible in AI search, and much of that is not bad content. It is access: crawlers turned away, or pages that read as blank to the machine even as they look perfect to you.

The access problem
~73%
Of businesses estimated effectively invisible in AI search, often from access and legibility gaps (industry compilation, SEJ).
Raw HTML
How major AI crawlers read. Most don't execute JavaScript, so script-painted content is invisible to them.
<3%
Amazon’s share of ChatGPT referral traffic (it blocks the crawlers) vs. 20%+ for Walmart and Etsy (Similarweb).

Four gates, and they are pass/fail

Between an AI crawler and your products sit four gates. Each is a yes-or-no, and a no at any gate ends the visit, with no error you will ever see.

Gate The question What fails it How to check
1. Access Are you letting the crawler in? A stray Disallow in robots.txt, often app- or theme-added Open yourstore.com/robots.txt
2. Render Can it read the page once inside? Content painted by JavaScript the crawler doesn't run View the page source; search for a product name
3. Reach Can it find every page? Orphans, dead URLs, a messy sitemap, broken canonicals Spot-check yourstore.com/sitemap.xml
4. Parse Can it understand what it found? Missing alt text, no structured data, ambiguous content Check images and schema on key pages
The front-door test. Most stores fail silently at gate 1 or 2 and never know it.
Four gates between an AI crawler and your products. AI CRAWLER robots.txt OPEN Can it get past robots.txt? AI crawler access OPEN Will the crawler even visit? JavaScript rendering PARTIAL Does the page render for a bot? page speed SHUT Does it finish before timeout? One shut gate ends the visit. It doesn't matter how good your products are if the crawler never reaches them. The front door is pass/fail, and most stores have never checked which gate is closed. This one times out before render. PASS PARTIAL BLOCKED VISIBILITY MESH THE FRONT DOOR TEST VM-C-P1 · r2.0 visibilitymesh.com
Four gates between an AI crawler and your products. A shut gate at any stage ends the visit, and you'd never see it happen.

Gate one in practice: the Amazon lesson

Your robots.txt is the bouncer, and one line decides everything. The cleanest proof is Amazon: it blocks the main ChatGPT crawlers, so its listings cannot surface in ChatGPT’s live shopping results, while Walmart and Etsy, which let them in, pull 20%+ of their referral traffic from ChatGPT. Same web, opposite outcomes, decided by one file. A well-prepared Shopify store can appear in an AI recommendation exactly where a giant’s listing cannot. Know which crawlers you are deciding about before you decide.

Gates two and three: read, then reach

The major AI crawlers largely do not run JavaScript. They read raw HTML. If your grid, price, or description only appears after a script, the crawler sees a blank where your products should be, and a slow page times it out mid-read. Even inside, it reaches pages by following links and your sitemap; redirect chains and contradictory canonicals teach it your store is unreliable, and anything in an image with no alt text is invisible. An optional clean signal is an llms.txt map.

A scan is a snapshot. Legibility drifts

Here is the part that decides whether this is a one-time fix or an ongoing discipline. Everything above can be true today and quietly false next month: a theme update overwrites a snippet, an app injects a duplicate, a redesign orphans a page. The machine-readable layer is the one humans never look at, so it regresses silently while the store still looks perfect. And gates swing shut on their own, an app update can add a disallow you never typed. That is why serious stores do not check once, they measure, fix, and re-measure. It is also why we re-scan our own store on a schedule, in public.

What we saw, and where it goes

The pattern through 2025 into 2026 was consistent: AI traffic to retail surged while a large share of stores stayed unreadable to the machines driving it. Growth on one side of the gate, locked doors on the other. We anticipate that gap closing fast, because the upside is now measurable and the fixes are known. The stores that open and clean their gates while most are still shut inherit a first-mover advantage that compounds.

Questions people actually ask

How do I know if AI crawlers can access my store?

Start with your robots.txt at yourstore.com/robots.txt to confirm the AI crawlers are not blocked, then check that your key content appears in the raw page source rather than only after JavaScript runs. A free scan checks these gates together and shows where a crawler would stop.

Why is Amazon a good example of crawler access?

Amazon blocks the main ChatGPT crawlers in its robots.txt, so its listings cannot appear in ChatGPT's live shopping results, while retailers that allow those crawlers draw a large share of referral traffic from ChatGPT. It shows one access decision can outweigh everything else.

Do AI crawlers read JavaScript?

Most of the major AI crawlers do not execute JavaScript; they read the raw HTML. If your products or key facts are painted onto the page by script, those crawlers can see a blank where your content should be, even though it renders fine for a human.

Is being blocked the same as ranking poorly?

No, it is worse. Ranking poorly means you were considered and placed low. Being blocked or unreadable means you were never considered at all, because the machine could not reach or parse the page. Access is the precondition for everything else.


See what a machine sees

You can't tell from your browser whether AI can read your store. You can find out in a few minutes. Run a free scan and see the exact layer the machine reads, and where you're losing the shortlist.

Run my free scan →

Sources: Search Engine Journal / industry compilation (2026) on AI-search invisibility; Similarweb (2025) on ChatGPT retail referral share and Amazon crawler blocking; multiple technical analyses (2026) on AI crawlers and JavaScript. Figures are third-party and current as of mid-2026; we publish our own benchmark data as our scan volume grows.