When you press run, the audit fetches the page you enter, usually your home page, and up to three more pages linked from it, picking pages such as about, services, contact, pricing and FAQ first. It reads them the way an AI engine's crawler would, looking for what lets an engine find your business, understand it, check it is who it says it is and quote it, and it asks Google for a speed reading on mobile.

Alongside that, it works out the question a customer would ask to find a business like yours, using your home page title, description, main heading and any address in your structured data, and searches for at most three competitors in your local market. That single question goes to Google AI Overviews, ChatGPT and Perplexity, all asked from one country: the country your web address ends in (.au, .nz, .uk, .ca, .ie, .sg or .za), or Australia for any other ending, such as .com. Google shows an AI Overview for some questions and not others, and when it shows none, that engine is left out of the count rather than counted against you.

The report opens with a single number, your AI Visibility Score out of 100, and its letter grade on the usual school scale: 90 and over is an A, then B, C and D in steps of ten, and anything below 60 an F. A line under the number says how many points stand between you and the next letter. Beside it sits one sentence drawn from the live answers, saying which engines named you and which competitor they named in your place.

Below that, every engine gets a row of its own. Each row says whether that engine named you, recommended a competitor and which one, or answered without naming you at all, and you can open the full answer it gave and read it for yourself. An engine that gave no answer this time says so plainly and is never counted against you. Three further rows, for engines the free scan does not ask, are marked as not tested instead of scored.

When the search turned up local competitors, a ranked list comes next: you and each of them, with one square per engine that answered, filled in when that engine named the business. It is the quickest way to see who currently holds the answer you want.

The score measures how ready your site is for AI engines, and the live answers set how high it can go. Being named matters most. A business none of them names cannot climb out of an F; each engine that does name you raises that limit, and only being named by all three lifts it completely. A page told to stay out of search results, or a missing foundation, pulls the limit down further. If none of the engines answers at all, the report tells you, and the score then covers your site alone.

Next, your site is rated in four areas: site access, answer readiness, authority and credibility, and business identity. Each one is marked Solid, Partial or Weak with the number of issues found in it, and nothing more: the report says where the problems sit without listing them. The individual issues, ranked by what they cost, are covered on the strategy session. Before that, a short block sets out what happens once AI names you, with figures from the AI agent we run on client websites.

The block on what this may be costing you is written by an AI model from your score and the competitor the engines named most, so treat any dollar figure in it as an illustration of scale, not a measurement of your sales.

There are things it cannot tell you. It asks one question, once, and the engines can answer the same question differently on another day. It counts you as named when an answer contains your web address or links to your site, so an answer that mentions your business by name alone is not counted.

Competitors come from a web search run by an AI model; if that search fails, a quicker step suggests them without browsing, and those suggestions are unverified. Pages are read before any scripts run, so content or structured data added by JavaScript is not seen. And with four pages at most and no access to your analytics, it cannot see your Google rankings, your traffic or the enquiries you already receive.