Free AI visibility audit — see your site the way AI sees it

AI assistants recommend products and providers in your category every day. Most brands have no strategy for it, and no measurement of how little of their site an AI can actually retrieve. The audit shows both, and what to fix first. It is free and requires nothing to be installed.

The finding the audit is built around

Across the web, an AI assistant retrieves roughly 15% of what a page publishes.

The average page publishes about 820 words. An AI assistant receives about 120 of them. The remaining 700 were written, reviewed, paid for and never arrive.

That figure needs its definition attached, because a similar-sounding number circulates with a different meaning. Adobe's AI Content Visibility Checker (May 2026) measured whether retail content is machine-readable in principle — cosmetics scored 63%, electronics 56%. Norg's 15% measures something narrower and more practical: the share of a page's words that actually arrive in a plain fetch. A page can be machine-readable in principle and still deliver almost nothing when an agent requests it.

Why so little arrives

Most of a modern page's content is assembled by JavaScript after load, hidden behind interaction — tabs, accordions, "read more" — or expressed visually in ways that carry no text. A human browser runs the scripts and clicks the tabs. A plain fetch does neither. What survives is the small portion present in the initial HTML.

Why this channel is worth measuring now

Signal	Figure	Source
Growth in AI-referred traffic to retail sites	393% year on year	Adobe Analytics, Q1 2026
Conversion of AI-referred traffic	42% better than every other traffic source	Adobe Analytics, March 2026
ChatGPT weekly active users	900 million	OpenAI via Reuters, February 2026
Claude monthly active users	245 million, up from 5 million in early 2025	Sensor Tower, mid-2026

The conversion figure is the one worth pausing on. AI-referred visitors convert better than any other source because the comparison work has already happened in the conversation before the click. They arrive having been told why you are the right choice.

What the audit actually checks

The audit fetches your site the way an AI system does and reports what comes back:

• Retrievable content ratio — how much of each page's published text survives a plain fetch.
• Structured data — whether JSON-LD is present, valid, and describes the entities that matter: your organisation, products, services, locations and people.
• Machine-readable endpoints — whether llms.txt, an MCP manifest and a traversable entity graph exist at all.
• Crawler access — whether robots.txt, a CDN-managed robots policy, or a WAF bot rule is blocking AI crawlers by name. This is the single most common invisible blocker: it produces no error signal anywhere, and every other health check reads green while nothing reaches the agent layer.
• Answer coverage — whether the questions buyers in your category actually ask have an answer anywhere on your domain.
• How AI currently describes you — what ChatGPT, Claude, Gemini and Perplexity say when asked about your business, and which sources they cite to say it.

The last item is usually the one that changes the conversation

Most brands have never seen the answer an AI system gives about them. When third-party sites are supplying that answer — review aggregators, directories, competitor comparison pages — the description is being written by parties with none of your history and no obligation to be accurate. B&D Garage Doors, an iconic Australian brand of 65+ years, found more than 70% of AI citations in its own category pointing away from its domain.

What happened to the brands who moved first

Brand	Result	Context
B&D (Dulux Group)	1,300% year-on-year growth in AI referrals	After 480+ product pages were made readable
Be Fit Food	+36% gross sales uplift	From AI-driven discovery, with no additional marketing spend
Selleys	AI visibility score 39 → 88	Measured before and after

Fuller detail on these and four other engagements is in the case studies.

Common questions

Is the audit really free?

Yes. There is no charge and nothing to install.

Do I have to change anything on my site to run it?

No. The audit fetches publicly available pages the same way an AI crawler would. It requires no code, no tag and no access to your systems.

What do I get back?

A report of what an AI system currently retrieves from your site, what structured data and machine-readable endpoints exist, whether anything is blocking AI crawlers, and how the major assistants currently describe your business — with an ordered list of what to fix first.

What if the audit finds nothing wrong?

Then you have a measured baseline and no work to do, which is a useful result. In practice the crawler-access check alone surfaces a blocker on a meaningful share of sites, because CDN and WAF rules commonly classify AI crawlers as bad bots without anyone having decided to block them.

Is this an SEO audit?

No, and the distinction matters. An SEO audit asks whether your pages can rank. This asks whether your facts can be retrieved and cited. A page can rank well and still deliver 120 of its 820 words to an agent — the two measures are independent.

What happens after the audit?

Nothing automatically. The report is yours. If you want the gaps closed, pricing is public and starts at $95/month for a single page.

Next step

Request the audit at norg.ai/ai-audit, or book a demo to see the platform against your own data.

---

Third-party figures are attributed to their sources above and were current as at the dates shown. Client results are specific to those engagements — see the individual case studies for measurement conditions.

Norg Pty Ltd — norg.ai