AI Search Visibility Audit
Find out whether AI assistants name your business when somebody asks a buying question in your category. Not a promise and not a package: a measurement, with the scope written down before you hand over a single detail.
Work directly with the person doing the work
Meet the team →The audit at a glance
AI Overview
An AI search visibility audit measures whether AI assistants name your business when somebody asks a buying question in your category. Loudachris runs 12 agreed prompts across 4 platforms (Google AI Overviews, ChatGPT, Google Gemini and Perplexity), 3 times each, and records all 144 responses. The report shows how often you were named, which businesses were named instead, and which URLs the platforms surfaced. The audit is free, the scope published on this page is the entire scope, and the result is a measurement rather than a ranking: model output varies between sessions, so frequency is reported and position is not.
A free audit with no stated scope is a lead form.
We went looking for an audit offer that said what it contained before asking for your details. We could not find one, so here is ours in full.
You cannot compare 2 offers that both say “free audit”
The first might be 144 recorded responses across 4 platforms. The second might be somebody typing your business name into an assistant once and screenshotting whatever came back. Both are advertised in the same 2 words, at the same price, behind the same button. Without a published scope you are choosing on branding, and branding is the one variable that tells you nothing at all about the work.
Unbounded scope always favours whoever is selling
If nothing was promised, nothing can fall short. An audit with no platform list, no prompt count and no described deliverable can be scaled down to whatever the seller feels like supplying and still technically be the thing you were offered. Writing the boundaries down first is what turns a gesture into an obligation, and it is the only reason the word audit carries weight on this page.
You cannot check a result you were not shown the shape of
A summary saying “your AI visibility is low” cannot be tested unless you can see the prompts behind it, how many times each one ran, and the answers that came back. That is why the raw log ships with the report. You should be able to run the same prompts yourself, or hand them to somebody who will, and hold what you get against what we recorded.
The Published Scope
Exactly what the free audit contains.
This is the whole thing. No larger version sits behind it and nothing here is withheld for a paid tier. If a row below is not what you need, the honest move is to say so before either of us spends time on it.
| Item | What we commit to |
|---|---|
| Platforms checked | 4: Google AI Overviews, ChatGPT, Google Gemini and Perplexity |
| Prompts per audit | 12, written as buying questions in your category and service area rather than as keyword strings |
| Who writes the prompts | You supply them, or we draft them and you approve the list in writing before anything runs |
| Runs per prompt | 3 on every platform, each in a separate session |
| Recorded responses | 144 in total (12 prompts x 4 platforms x 3 runs) |
| Competitors compared | 3 businesses you nominate, plus any business the platforms named more often than yours |
| Deliverable | A written report in 6 sections, plus the raw response log as a spreadsheet |
| Turnaround | Within 5 business days of your completed intake details |
| Cost | Free. No card, no obligation, nothing attached to the report |
| What we need from you | Business name, website, service area, 3 competitor names, and either your 12 prompts or your approval of ours |
| What it is not | A ranking report, a guarantee, or a subscription |
The only price on this page is our published SEO pricing, from $1,500+GST/month, and that applies only if you decide you want work done after reading the report. The audit itself carries no figure because it carries no charge.
Method
How the measurement is taken.
4 steps, in this order, with the first one locked in writing before the second one starts.
1
Agree the prompts
You send the business name, website, service area and 3 competitor names. We draft 12 buying questions from that, or you write your own. Nothing runs until the list is signed off, because a prompt list chosen afterwards can be chosen to flatter the result.
2
Run them, 3 times each
Every prompt goes to Google AI Overviews, ChatGPT, Google Gemini and Perplexity, 3 times, in separate sessions. That is 144 recorded responses. The repetition is not padding: it is the only defence against a single lucky or unlucky answer.
3
Record the output verbatim
Each response is captured as it came back, with the platform, date and run number attached, along with which businesses were named and which URLs were surfaced. Nothing is summarised on the way in, so the log stays checkable.
4
Trace the gaps to causes
Where you were named less often than a nominated competitor, we look at the observable difference: pages that exist or do not, business data that agrees or disagrees with itself, review depth, profile completeness. Observable causes only.
The variance disclosure, up front.
Generative systems do not return the same answer to the same question. Ask twice and you can get 2 different lists of businesses, 2 different sets of sources and 2 different tones. Session history, the location a request appears to come from, and the model version serving that request all move the output. Anyone showing you a single screenshot as proof of AI visibility, flattering or damning, is showing you 1 roll of a die.
Running each prompt 3 times reduces that noise. It does not remove it. So the audit reports how many of 3 runs named you, on which platform, on a stated date, and stops there. It does not convert that into a score out of 100, a visibility percentage or a position, because none of those exist to be measured and inventing one would make the report feel more precise than the underlying data is.
A repeat run at a later date is the only way a single audit becomes a trend. We are happy to re-run the same 12 prompts on request so you have a second reading against the first, and the scope of that repeat is identical to the scope above.
Limits
What this audit will not tell you.
Published for the same reason as the scope. A diagnostic that claims no limits is not a diagnostic, it is a sales asset.
✕Why a model chose a source
No platform publishes its selection or weighting criteria. Any account of why one business was named and another was not is inference. The audit reports what was named and what was surfaced, and does not narrate a decision nobody outside the platform can see.
✕That anything will change
Measuring a thing does not move it. The audit tells you whether you appear to have a problem and what the observable causes look like. It cannot commit to a different result later, and no honest reading of the data would let it.
✕A ranking, because there is not one
AI answers have no fixed order and no stable slots, so there is nothing to be number 1 in. Frequency across a stated number of runs is the strongest claim the data supports, so frequency is what the report contains.
✕What it is worth in revenue
A mention is not a click, a click is not an enquiry, and an enquiry is not a job. The audit measures visibility inside a specific set of answers, not demand, and it will not model a dollar figure out of that.
✕A trend, from a single reading
A single audit is a baseline taken on a stated date. Direction of travel needs the same prompts run again later, and the report says plainly that a first reading is a starting point rather than a movement.
✕A list of AI-only fixes
There is no separate AI channel to optimise for, and no AI-specific technique the report will hand you. Where a finding is actionable, the action is ordinary search work. That is the whole argument this part of the site makes.
The Deliverable
The structure of the report you receive.
6 sections and a raw log, in this order. Every audit report uses this structure, so you know the shape of the document before you ask for one.
Scope confirmation
The exact 12 prompts, the 4 platforms, the dates and the run count, restated at the front of the report. This is the section that lets you repeat the audit yourself, or hand it to somebody else to repeat.
Visibility summary
For every prompt, on every platform, how many of the 3 runs named your business. Reported as a frequency out of 3, never as a position, because there is no position to report.
Competitor comparison
The same frequency table for the 3 businesses you nominated, plus any business that turned up more often than you did. Names you gave us, and names the platforms gave us.
Citation list
Which URLs each platform surfaced or linked when it answered. Yours, your competitors', and the directories, forums and mastheads sitting in between. Observable output, not an inference about weighting.
Entity and data check
What the platforms state as fact about your business (services, trading name, location, hours, phone) set against what your own website and Google Business Profile say. Disagreements are flagged with the source of each version.
Findings, ranked by whether they are worth acting on
Each gap traced back to an ordinary cause: a page that does not exist, business data that disagrees with itself, a thin review profile, a profile with empty fields. Some findings are marked not worth acting on, and we say so plainly.
The raw response log
All 144 recorded responses, verbatim, with the platform, date and run number against each one. Supplied as a spreadsheet so you can read the evidence instead of taking a summary on trust.
Findings stay with the business that asked for them. No sample results, extracts, or anonymised “typical” numbers from previous audits appear on this page or anywhere else on this site, because a report is only worth having if the business it belongs to controls where it goes.
Ruled Out
Tactics the report will never recommend.
These are the line items we see turn up in AI visibility proposals, each one set against what the published guidance actually says. We criticise the tactic, never the firm selling it.
| Sold as | Status | Source |
|---|---|---|
| llms.txt file | Google's AI features guidance states it is not needed | AI features and your website |
| AI-specific schema vocabulary | No such schema.org vocabulary exists for generative AI | AI features and your website |
| Content chunking "for AI" | No requirement to break content into small pieces | AI features and your website |
| Writing specifically for AI | Not required. Write for people | AI features and your website |
| Speakable schema as an AI lever | Limited availability, news publishers only | Speakable structured data documentation |
| HowTo rich results | Retired from Google Search | Search Central, HowTo and FAQ changes |
| Guaranteed AI citations | Citation selection is probabilistic and varies by phrasing, session and model version | AI features and your website |
Sources: Google Search Central, AI features and your website for the llms.txt, schema, chunking, writing and citation rows, the speakable structured data documentation for the speakable row, and the Search Central post retiring HowTo rich results for the HowTo row. Checked 9 August 2026. Where a claim on this page is not sourced, it is our own first-hand observation and is written as one rather than dressed up as data.
The Position
The audit is new. What it measures is not.
The signals that decide whether an assistant can confidently name your business are the signals that decided whether a search engine could: pages that exist and say something specific, business data that agrees with itself everywhere it appears, a review profile with depth and recency, and a complete Google Business Profile. In other words, this is SEO, and it belongs inside an SEO retainer rather than in a second budget line with a newer acronym on it.
What is genuinely new is the measurement. Nobody had to check whether a machine would recite their business back to a customer until recently, and that is a fair thing to want checked. So the diagnostic is the product here, given away with the scope published, while the remedy stays inside ordinary search engine optimisation. The full argument, including what the labels mean and where they came from, sits on the AI search visibility hub, with the term-by-term breakdown on answer engine optimisation and the evidence review on what actually works.
Afterwards
3 honest outcomes, and one of them is “change nothing”.
The platforms already name you
This is a real outcome, not a token entry on the list to make the page look balanced. If the log shows you named consistently across the runs, the recommendation is to keep doing what you are doing. There is no upsell attached to that answer, and no reason to invent one.
There are gaps you can close yourself
Some findings are small jobs: an old phone number sitting in a directory, a service you offer that no page on your site mentions, opening hours that contradict your profile. The report names them, tells you where they are, and you are welcome to fix every one of them without us.
The gaps need ongoing search work
Thin content across a whole service range, a review profile with nothing recent in it, or business data scattered across the web in 4 versions are not small jobs. That is an SEO engagement, and ours starts from $1,500+GST/month with its scope written down the same way this audit is.
No path from this page ends in a surprise invoice. The audit is free because a measurement held back from the business it describes is no use to anybody, and because we would rather earn the next conversation with a document you can check than with a claim you cannot. Pricing detail sits on the SEO cost guide, what a retainer covers sits on the SEO services page, and the local version of this work sits on local SEO.
Questions about the AI search visibility audit.
What does the AI search visibility audit cost?
Why publish the scope when almost nobody else does?
Which AI platforms do you check?
How many prompts do you run, and who writes them?
Why run the same prompt 3 times?
What does the report actually contain?
Can you tell me why a model named my competitor instead of me?
Will the audit tell me how to rank in AI answers?
Is this the same thing as AEO or GEO?
What happens after the audit? Do I have to buy anything?
Do I need an llms.txt file, AI-specific schema, or content broken into chunks?
Do you audit businesses outside Adelaide?
Who wrote this
Chris Lourenco, Director, Loudachris Digital Marketing
Loudachris Digital Marketing (ABN 73 630 143 190) is a Google marketing practice at 21-22 Greenhill Road, Wayville SA 5034, working on SEO, Google Ads and the websites that support them. The audit described on this page is run by the same person who wrote the page, and the scope above is the version we are held to. Where a claim here is sourced, the source is linked with the date it was read. Where it is a judgement, it is written as one.
Last reviewed: 9 August 2026
More on AI search visibility
The rest of the argument, and where the audit sits inside it. Start at the AI search visibility hub if you want the whole thing in order.
AI search agency
What to ask anybody selling an AI visibility retainer, and which answers should end the conversation.
Generative engine optimisation
Where the GEO label came from, what it claims to cover, and how much of it already had a name.
What actually works
The short list that survives scrutiny, separated from the long list that does not.
AI search in Adelaide
The same question asked from a South Australian address, and why location changes the prompt.
Find out whether you have a problem worth solving.
Send the business name, the website, the service area and 3 competitors. We will draft the 12 prompts for your approval, run them, and hand back the report with the raw log attached. Free, and the scope above is the scope.
Call 0403 454 199
Prefer the argument before the measurement? Read the AI search visibility hub or the full SEO services page.