Free tool · about 2 minutes · no signup to run
Can an AI assistant find your store, read it, and recommend it?
An AI visibility audit measures two separate things: whether AI assistants are able to reach and parse a store, and whether they go on to name it once a shopper asks for a recommendation. The ForgeCRO AI visibility audit requests your pages under 13 published agent user-agents, then puts buying questions to the assistants themselves and scores both halves out of 100.
Run it here
Free · about two minutes
It is hosted at audit.forgecro.com. If nothing shows after a few seconds, use the audit in its own tab.
Reads only public pages. Your store data, admin and customers are never touched.
- 01
Enter your store address
The audit requests your pages the way AI assistants and shopping agents do. There is nothing to install and no login to share.
- 02
See where you stand
A 0 to 100 score split by pillar, each finding shown with its evidence, and any rival an assistant suggested in your place.
- 03
Book 30 minutes
Thirty minutes on screen with the engineer who reads the results. You leave with a fix order, whatever you decide next.
Rather write?info@forgecro.comOr pick a 30-minute slot. Every brief gets a written reply from the team within one working day.
A brand of Aqsa Shahzad Official LLC
What it reads
What an assistant can and cannot use.
What it reads
- Product feed and GTIN / MPN coverage on every variant
- Product, Offer and Organization markup, and whether it actually says anything useful
- robots.txt, llms.txt and the agent user-agents you block
- Live answers from the assistants for your brand and your category
- Sitemap, canonical and crawl health an agent depends on
What you get
- A score out of 100, by pillar, with what each point is worth
- Every finding with its evidence and the points it recovers
- The competitor named, if an assistant recommends one instead of you
- The fix order — what moves the score first
- The report by email, to a work address
The audit queries
- ChatGPT
- Claude
- Perplexity
- Gemini
- Search Console
- Shopify
Watch it run
The whole thing, end to end.
The same tool as above, recorded on a real store. Sound is off until you ask for it.
AI visibility questions
What people ask before they run it.
Is the AI visibility audit free?
Yes, and it runs without an account. You see the score and the findings on screen; a work email is only needed if you want the full report sent to you.
How long does a single run take?
Roughly two minutes, and most of it is spent waiting on the assistants. The audit poses genuine buying questions to them instead of inferring an answer from your markup.
Which AI assistants does it check?
Thirteen published agent user-agents are requested one by one. They belong to the crawlers operated by Anthropic, Perplexity, OpenAI, Microsoft, Apple and Amazon, plus Google AI Mode. Separately, buying questions go to each assistant that exposes a live-search endpoint.
Does it need access to my store?
None. The audit is served exactly what an assistant would be served — your public pages — so there is nothing to connect and no permission to grant.
What does a low score usually mean?
In most stores it is one of three things: a CDN rule that turns agents away while robots.txt says they are welcome, product data with no GTIN or MPN on the variants, or policy pages an assistant cannot verify. The report separates them, because the fixes are not the same.
Why does robots.txt say yes but the audit says blocked?
Because the two are answered by different systems. robots.txt is a file you publish; the block is usually applied at the edge by a bot-protection rule that never reads it. The audit records the status code each agent actually received.
Who reviews the result?
The score is produced by the tool. If you send the result in, a senior engineer at ForgeCRO reads it and replies in writing, usually within one working day.
What happens to my store data?
Nothing is stored beyond the audit record. Only public URLs are requested, and no customer, order or admin data is ever touched.
If it found something
An assistant that cannot read you will recommend someone it can.
Run it free, then walk through what it found on a 30-minute call. You leave with a priority list either way.