AI visibility

Website analysis for AI
what’s reviewed and why it matters

More and more people ask an AI assistant instead of a search engine. For your website to appear in those answers, the AI has to be able to get in, read the content and understand what it’s about. Wakaris reviews it with 10 checks and explains every finding in plain language.

  • 10 checks
  • 5 possible findings
  • 5 guides
What’s reviewed

What Wakaris checks
about your website and AI

That’s 10 checks grouped into 5 blocks. Each block is a finding that can show up in your report.

Finding01/ 05

Limits1 check

AI crawler access blocked

For an AI assistant to mention you, it has to have been able to read you. It’s the precondition for everything else, and it’s the one most websites fail without knowing: many security plugins and CDNs block these crawlers by default.

What it checks

  • AI crawler accessFor each AI crawler in the catalog, evaluates whether the applicable rules allow it to access the URL. No finding if all are allowed; limits if some are and some aren’t; blocks if all the search and retrieval crawlers evaluated are blocked.
Read the guide: AI crawler access blocked 

Finding02/ 05

Limits1 check

Content without JavaScript

Google runs JavaScript, but many others don’t: most AI crawlers, link preview generators and content readers read the HTML as it arrives and leave.

What it checks

  • Content readable without JavaScriptCompares the HTML received with the rendered result. No finding if the main content already has visible text before JavaScript runs; blocks if it has no text and the text appears later; inconclusive if there’s no text content afterwards either.
Read the guide: Content without JavaScript 

Finding03/ 05

Limits5 checks

Structuring for citability

An assistant that summarizes your content needs to know who it’s from and when it’s from to be able to cite it. If that data isn’t structured, it may use your text without attribution — or attribute it wrongly.

What it checks

  • Structured authorshipIf there’s an Article, BlogPosting or NewsArticle type, checks the author property. No finding if it’s present and not empty; blocks if it’s missing or empty; inconclusive if none of those types are present.
  • Structured publication dateFor those same types, checks datePublished. No finding if the date is present and valid; blocks if it’s missing or the format is invalid.
  • Consistent modified dateIf dateModified exists, validates the format and the time relationship. No finding if it’s valid and not earlier than the publication date; blocks if it’s invalid or earlier; inconclusive if it isn’t declared.
  • Structured FAQIf there’s an FAQPage, checks each Question. No finding if all have a non-empty name and accepted answer; blocks if at least one is incomplete.
  • Structured Q&AIf there’s a QAPage, validates the required question and answer fields. No finding with the full structure; blocks if any required element is missing.
Read the guide: Structuring for citability 

Finding04/ 05

Affects2 checks

llms txt file

llms.txt is a recent proposal: a Markdown file that sums up what your site is about and which pages are worth visiting, designed so an assistant can understand your website without crawling all of it.

What it checks

  • llms.txt fileRequests /llms.txt. No finding with a 200; inconclusive with a 404 or 410, because it isn’t mandatory; blocks if there’s a reference or implementation but it returns 401, 403, 5xx, a loop or an invalid response.
  • llms.txt formatIf it exists, validates its structure against the defined format: valid Markdown, a top-level main heading and a parseable structure of links and sections. No finding if there are no problems; blocks if there’s at least one structural problem.
Read the guide: llms txt file 

Finding05/ 05

Affects1 check

Semantic HTML use

A `<div>` says nothing: it’s a box. An `<article>`, a `<nav>` or a `<footer>` say what role that block plays, and that’s what lets a machine separate content from decoration without guessing.

What it checks

  • Semantic container ratioCalculates semantic / (semantic + div) × 100, counting header, nav, main, section, article, aside, footer, figure, figcaption, details and summary as semantic. No finding above 75%; limits between 25% and 75%; blocks at 25% or below.
Read the guide: Semantic HTML use 

The label tells you how much each finding weighs: Limits important,Affects could be better.

Why it matters

Why it matters that AI
understands your website

An assistant can only talk about you if it has been able to read you first.

  • A new way to search

    Someone who asks an assistant gets a ready-written answer, not a list of links. Either your website is among what the AI has read, or it doesn’t appear.

  • Unread means invisible

    Some websites shut out AI crawlers without knowing it, or only show their content after running code that those crawlers don’t run.

  • Get quoted accurately

    Clear, well-organised content is easier to summarise without mistakes. The better your page is understood, the better what you do gets told.

To understand it

Artificial intelligence guides
to understand each finding

Each guide explains one finding from this area in plain language: what it is, how it’s measured, why it matters to you and how to fix it. The technical part is kept separate, for whoever looks after your website.

5guides on artificial intelligence

See the 5 guides
Frequently asked questions

Common questions
on artificial intelligence

Does Wakaris tell me whether ChatGPT talks about my company?

No. Wakaris doesn’t ask assistants about you or measure how often they mention you. It reviews what depends on your website: that their crawlers can get in, that the content can be read and that it’s organised in a way that makes sense.

Can I stop AI from using my content?

You can ask it not to: the robots.txt file states which crawlers may come in and which may not, and the main ones respect it. The analysis tells you how it’s set up right now, so that it’s your decision and not just the way it came by default.

What is the llms.txt file?

A text file, at the root of your website, that sums up for AI assistants what’s on it and where the important things are. It’s a recent proposal and not everyone uses it yet, but it takes little effort to have one.

If I already look after my SEO, do I need anything else?

Much of the work is shared: a clear, well-structured website serves both purposes. But there are differences, such as access for AI crawlers or content that only appears once code has run. That’s why they’re separate areas.

Who has to fix whatever comes up?

Whoever looks after your website. Several findings are changes to settings or structure, without redoing the content. Each one comes explained, with the evidence that proves it and a ready-made prompt to fix it with an AI, and you can share it with a link.

How much does this analysis cost?

Nothing. Analysing a page is free, asks for no sign-up or card and gives you the result in a few seconds. This area is reviewed alongside the other eight in the same analysis.