Works for online stores and company sites. All you need is the address: no access to your accounts, no code on your site, no call with a salesperson.
The checker reads only the public site and its files, the way a search engine bot does. It buys nothing, submits no forms and collects no visitor data.
Model bots do not click and usually do not run scripts. The checker reads your site the same way they do, then compares it with what a person sees in the browser.
We fetch the plain HTML, without running JavaScript, and count how much text, how many headings and how much structured data is left in it for a model.
We read the rules for bots, the file for language models and the sitemap, including the Content-Signal lines, which say what the content may be used for.
We request the site under the name of each bot separately and compare the response with the one a browser gets. That is how a firewall that lets people in and turns models away shows up.
From the sitemap we take up to three subpages and check whether their content and schema reach the bot the same way the homepage does.
We check whether the site gives models clean content in Markdown, and in a cloud browser we compare the text before and after scripts run.
A score across five areas, a table of bots and a list of blockers from the most serious one, each with proof from the server's response.
Each area has its own weight in a score of up to 100 points. Bot access weighs the most, because a model will not cite a site that a bot cannot enter, whatever its content.
Whether robots.txt, the firewall and the headers let in the bots of model search engines and those that open a link at a user's request, and whether the page itself tells them noindex.
How much of the homepage and subpage content is visible in the HTML alone, before scripts run, and whether the headings, title and description add up to a clear whole.
Company data in Organization, schema matched to the page type (product, category, article), valid JSON-LD, and whether the data sits in the HTML rather than only in a script.
Questions with answers, content split into sections, author and date on guides, and contact details that let a model recognise a real company.
The llms.txt file, a Markdown version of pages, content-use signals in robots.txt, and content freshness based on dates in the sitemap and on the pages.
The score, the table of bots and the list of blockers are visible without leaving any details. An email is needed only for the repair instructions and the PDF report, which you can pass to a developer or an agency.
Models see the site only partly, so they cite it less often than they could.
ChatGPT, Perplexity, Gemini and Google AI Overviews answer customers with a sentence put together from several pages. A site that a bot cannot read will not make it into that answer, even when it is the best source in its category. Most often a small thing stops the bot: one line in robots.txt, a firewall set against bots, or content that only loads through a script.
The analytics scanner opens the site in a real browser and checks the tag manager, GA4, Meta and Google Ads pixels, Consent Mode and server-side. Every issue comes with evidence, and leaving an email unlocks the fix instructions.
The scan checks whether models can read the site: whether bots get in, whether content and structured data reach them without JavaScript, and whether there is anything to cite. That is the condition without which a model will not name the brand. Whether it names the brand, and how often next to competitors, is measured by the AI brand visibility audit, in which we ask the models a set of questions from your category many times.
A search bot (OAI-SearchBot, Claude-SearchBot, PerplexityBot) reads a site so that the model can cite it and link to it in an answer. A training bot (GPTBot, ClaudeBot, CCBot) collects text to train the next versions of the model. Blocking training is a company decision and the checker does not count it as an error. Blocking a model's search bot cuts the site off from citations in that model.
It is not an official standard and Google does not read it. Coding assistants and some agents use it, and it is cheap to prepare, so the checker counts it as a small part of the score. Far more weight sits on whether bots can enter the site and whether the content is visible without JavaScript.
The bots of model search engines usually fetch only the HTML and do not run scripts. If a product description, a price or category content appears only after JavaScript loads, the bot sees an empty page. The checker compares the text from the HTML alone with what the browser shows after the scripts run.
No. It reads only public pages and files, the way a search engine bot does: the homepage, up to three subpages, robots.txt, llms.txt and the sitemap. It does not log in, submit forms or collect visitor data.
The email unlocks the repair instructions for each blocker and the PDF report. We use it only to contact you about the scan result, in line with the privacy policy. The score, the table of bots and the list of blockers are visible without it.
Nothing. The scan, the result and the PDF report are free. Only the fix is paid, if you decide to hire us for it, and its price is shown next to the services below the result.
The call is led by a new business specialist. When we get into the details of an account, the specialist for that channel joins in. We reply within one business day.
Before the meeting we review your website, your visibility and what your campaigns show from the outside. We will not open with “so, what does your company do?”.
The reply comes from the same new business specialist who will run the call. No qualification form and no call from an unknown number.
We go through your numbers and your questions. If it turns out we are not the right fit, we will tell you straight away.
