AI Visibility Audit | Umer Qureshi

Most businesses have never seen what AI assistants say about them. Not a guess, not a sample of two questions typed into a chatbot once, but a recorded answer to every question a buyer would actually ask, across every major assistant, with the citations captured.

This audit produces that document. It is deliberately a fixed piece of work with a defined end rather than a retainer, because the finding usually changes what a business wants to do next, and it is unreasonable to ask anyone to commit to a programme before they have seen the evidence.

A meaningful number of these audits end with a recommendation not to start a large content programme. That is a legitimate outcome.

What the audit covers

Twenty to forty questions, written to match how your buyers actually phrase things, run across ChatGPT, Perplexity, Gemini, Copilot and Google AI Overviews. Every answer recorded verbatim with its citations.

For each question the report records whether you were named, what you were named for, whether the description was accurate, which competitors appeared, and which sources the answer was built from. That last column is usually the most useful, because it shows exactly what a model considers a credible source in your category.

The diagnosis, not just the score

A number saying you appear in eleven percent of answers is not useful on its own. The audit explains the mechanism behind it.

Typical findings: the facts a buyer needs live in a PDF or an image that nothing can read. The site describes itself three different ways across three pages, so no description is confident enough to quote. Competitors are being cited from a directory the business has never registered with. Or the pages are fine and the problem is that nothing outside the site corroborates them.

Each of those has a different fix and a different cost, which is why the diagnosis matters more than the score.

The output

You receive a written report with the recorded answers as an appendix, a competitor comparison across the same question set, the diagnosis, and a prioritised fix list ordered by expected return against effort.

The fix list is written so another supplier could execute it. It is not structured to make hiring me the only option, and where an item is better done in-house or by your existing agency the report says so.

What happens afterwards

Three outcomes are common. The fix list is small and your team executes it, which is the best result and the one the report is written to enable. The fixes are structural and you want them done as an engagement, which is when the AEO or AI SEO service applies. Or the finding is that visibility is not the constraint and the money is better spent elsewhere.

The audit is priced and scoped as a standalone piece of work precisely so all three of those endings are available.

Why this is scoped as a fixed piece of work

Retainers create an incentive to find ongoing work. A fixed-scope audit does not, which is the point. The report is written to be actionable by anyone, including your existing agency, and where the fixes are small enough to do in-house it says so plainly.

That also makes it a reasonable first engagement for a business that has not worked with me before. You get a complete, standalone deliverable and a clear view of how I work, without committing to a programme on the strength of a proposal.

What tends to surprise people

Two findings come up repeatedly and both tend to land harder than the headline number.

The first is being described inaccurately rather than being absent. A business that repositioned two years ago frequently finds assistants still describing the old offer, because the old description is what the wider web still repeats. That is a corroboration problem and it does not fix itself.

The second is discovering which sources assistants actually trust in your category. It is often a trade directory or a review platform nobody internally considered important, and getting listed there properly is sometimes the single highest-return item on the whole fix list.

Getting started

The audit needs one conversation to agree the question set, and roughly one to two weeks to run and write up.

You get the report, the recorded answers as an appendix, the competitor comparison, and a fix list ordered by return against effort. The fix list is written so anyone can execute it, including your existing agency.

There is no obligation to continue afterwards, and a fair share of these end with the fixes being done in-house. That is a successful outcome, not a lost sale.

What the engagement includes

A recorded baseline. Twenty to forty buyer questions run across five assistants, with every answer and citation captured verbatim as an appendix.
A competitor comparison. Who is named instead of you, on which questions, and which sources their answers were built from.
A written diagnosis. The mechanism behind the result, not a score. Why the answers look the way they do and what is actually causing it.
A prioritised fix list. Ordered by return against effort, written so any competent supplier could execute it.
A re-measurement baseline. The question set and recording format, handed over, so you can re-run it yourself later.

How the work runs

Build the question set. Twenty to forty questions phrased the way your buyers phrase them, covering the whole buying sequence.
Run and record. Every question across five assistants, answers and citations captured verbatim.
Compare. The same set analysed for competitor presence and source patterns.
Diagnose. Establish the mechanism behind the result rather than reporting the score.
Prioritise. A fix list ordered by return against effort, executable by anyone.

Proof

This audit is the usual starting point before the answer engine optimization or AI SEO engagements.

Talk about your situation

The first conversation is short and mostly questions. Get in touch and tell me what you are trying to fix. Or see how this is priced.

Frequently Asked Questions (FAQs)

How do I check if AI mentions my brand?

Ask each major assistant the questions your buyers would ask, and record the answers and citations rather than reading them once. The discipline that makes it meaningful is a fixed question set run repeatedly, because model output varies between identical prompts and a single test tells you almost nothing.

What does an AI visibility audit include?

A recorded baseline across five assistants, a competitor comparison on the same questions, a written diagnosis of why the answers look the way they do, and a prioritised fix list. The recorded answers are supplied as an appendix so the conclusions can be checked.

How long does the audit take?

Usually one to two weeks from the question set being agreed. Most of that is running and recording, which is deliberately not rushed, since the value of the baseline depends on it being complete rather than sampled.

Do I have to buy an ongoing service afterwards?

No, and the report is written to make that unnecessary. The fix list is executable by your own team or your existing agency. A fair share of these audits conclude that the fixes are small enough to do in-house, or that AI visibility is not the binding constraint on the business.

Is this different from a normal SEO audit?

Yes. A conventional SEO audit examines crawlability, structure and rankings on your own site. This examines what external AI systems currently say about you and why. There is overlap in the fixes, but the evidence base is entirely different: recorded assistant answers rather than crawl data.

Can you audit a competitor instead of us?

Yes, and it is a useful exercise. The same method run against a competitor shows which questions they own, which sources their answers are built from, and where their coverage is thin. That is often the fastest way to find the gap worth attacking, and it is frequently commissioned alongside your own baseline.