Free · No call required
AI visibility audit: find out whether answer engines can read you at all
I test your site against the four engines your buyers actually use, check whether their crawlers can reach and parse you, and send back a one-page readiness score with the three fixes I would make first. Three working days. No obligation, and no call unless you want one.
- Tested across ChatGPT, Gemini, Perplexity and Claude
- Arabic prompts included where your market warrants it
- Written by me, not generated by a tool
Thank you for your response. ✨
What it is
What an AI visibility audit actually checks
Three different things get sold under similar names, and they are not interchangeable. It is worth being precise, because the audit only makes sense once you know which layer you are failing at.
| What it asks | Who decides | |
|---|---|---|
| SEO | Can a crawler reach, parse and rank this page? | Google’s index |
| AEO | Is there a clean, self-contained answer on the page? | The page’s own structure |
| GEO | When a model answers this question, is it us it quotes and attributes? | The model’s retrieval and citation choice |
An AI visibility audit covers all three, in that order, because they stack. A site that fails the first question cannot pass the third — generative engines overwhelmingly read pages that conventional retrieval already found. That is why this is an audit rather than a tactics list: the answer usually is not “write more”, it is “you are invisible three layers earlier than you think”.
Scope
What I check
Reachability
Whether GPTBot, ClaudeBot, PerplexityBot and Google-Extended are allowed to crawl you at all. This is blocked by accident on a surprising share of sites — usually a robots rule inherited from a template or a CDN default nobody revisited.
You get: a yes/no per crawler, and the exact directive to change
Renderability
Whether your content exists in the served HTML or only appears after JavaScript runs. Retrieval pipelines mostly see the served HTML, so this single check explains more invisible sites than anything else.
You get: a served-versus-rendered comparison across your key templates
Answer structure
Whether a model can lift a self-contained answer out of your pages, or whether the answer is buried, hedged into uselessness, or split across three paragraphs that need each other for context.
You get: the passages that are quotable, and the ones that are not, with the reason
Entity clarity
Whether your Organization and Person facts agree across your site, LinkedIn, directories and profiles. A model that cannot resolve who you are will not attribute a claim to you.
You get: a list of every inconsistency found, and where it lives
Prompt-set visibility
Up to ten real buyer questions, tested across four engines, recording where you appear, where a competitor appears instead, and what is actually being cited.
You get: the tested prompt set and a baseline you keep and can re-run
Scoring
The readiness score
Every audit ends with a single number, so the conversation with your board is short. The bands mean specific things, not vibes:
| Score | What it means | What to do next |
|---|---|---|
| 85–100 | Retrievable and quotable. You appear for most of your prompt set. | Refine coverage and monitor. No structural work needed. |
| 70–84 | Readable, but losing citations to better-structured competitors. | Answer-first restructuring and cannibalisation control. |
| 50–69 | Reachable but not quotable. Ranking without being cited. | Content restructuring and entity clarity, in that order. |
| Below 50 | Effectively invisible to answer engines. | Stop all content spend. Fix crawl and rendering first. |
Most sites I audit for the first time land between 40 and 65. That is not a sales line — it is what happens when a site was built for humans and conventional search, and nobody has revisited the assumptions since answer engines started reading the web.
Deliverable
What you receive
- A one-page readiness score with the band and what it means for you
- Crawler access status for GPTBot, ClaudeBot, PerplexityBot and Google-Extended, with the fix for any that are blocked
- A served-versus-rendered check across your key templates
- Your prompt set, tested across four engines, with a baseline you keep
- Who is being cited instead of you, for the prompts where you do not appear
- The entity inconsistencies found across your site and off-site profiles
- The three fixes I would make first, in order, with the reason for the order
- An honest note on whether this is worth paying anyone to fix — including me
It arrives as a document, not a dashboard login, and it is written by me rather than generated. If the honest answer is that your site is fine and you should spend the money elsewhere, that is what it will say.
How it runs
Four stages, in this order
01 · Day 0
You send the form
Website, sector, market, and the questions a buyer should be asking an AI to find you. Two minutes.
02 · Days 1–2
I test
Crawler access, rendering, answer structure and entity consistency, then your prompt set across four engines.
03 · Day 3
You get the audit
Score, findings, baseline prompt set and three prioritised fixes. By email.
04 · After
Your call
Implement it yourself, hand it to your team or agency, or talk to me about the work. No follow-up sequence either way.
Fit
Who should request one
- Pharmaceutical and healthcare marketers whose content passes med-legal review but never gets quoted.
- B2B and industrial brands with long consideration cycles, where buyers now start at an AI assistant rather than a search box.
- Anyone who has just migrated and cannot tell whether the drop is ranking, rendering or both.
- Brands running Arabic and English who suspect the Arabic side is invisible but have no way to prove it.
- Marketing leads being sold GEO retainers who want an independent read before signing one.
If you are about to commission AI-visibility work from anyone, get this first. It costs nothing and it tells you whether the proposal in front of you is addressing your actual problem.
Questions
Audit questions
ChatGPT, Gemini, Perplexity and Claude. Those four cover the overwhelming majority of assistant usage among the buyers my clients sell to. I also check the crawler access for each, which is a separate question from whether you currently appear.
Up to ten, built from the buyer questions you give me on the form plus any obvious ones you have missed. Ten is enough to establish a baseline and see a pattern; more than that becomes a paid engagement rather than a free snapshot.
Yes, where your market warrants it, and the prompts are written per market rather than translated — an Egyptian buyer and a Saudi buyer phrase the same question differently. Arabic visibility is usually far weaker than clients expect, and the competition for it is far thinner.
It is free and there is no catch beyond the obvious one: I do this because a percentage of people who read an honest audit decide to hire me. If the audit says your site is fine, it will say so, and you will not hear from me again unless you get in touch.
Three working days from the form. My week runs Sunday to Thursday, Cairo time.
Whatever you want. Implement it yourself, hand it to your team or agency, or talk to me about the work. There is no follow-up sequence and no call booked on your behalf.
Not for the snapshot — everything in it is testable from the outside, which is rather the point, since that is all an answer engine gets too. A full technical SEO engagement does need Search Console and ideally server logs.
Those run a crawler and print a checklist. This tests your actual prompt set against four live models and tells you who is being cited instead of you, which no crawler can see. It is also read and written by a person, which is why it takes three days rather than thirty seconds.
Start with an audit, not a retainer
Fixed scope, fixed price. You get the findings whether or not we work together afterwards.
Related:
Technical SEO & site architecture
Generative Engine Optimization
All six services