An AI visibility audit should show its work.
A model answer is an observation, not a stable ranking. Preserve the question, conditions, answer, citations, public facts, and repair—or do not call the result evidence.
No prompt score. No recommendation guarantee. Nothing entered on this page is uploaded or saved.
- 01ASK
Exact buying question
Keep the wording, platform, location, account state, and time with the observation.
- 02RECORD
Answer as observed
Name the business, wording, position, omissions, and uncertainty without turning one answer into a rate.
- 03TRACE
Sources and citations
Open every cited page. If the answer gives no source, record that instead of inventing one.
- 04CHECK
Public business facts
Compare the answer with the live site, Business Profile, structured data, reviews, and enquiry path.
- 05REPAIR
Reproducible next action
Fix the public contradiction or missing source that another person can verify after publication.
Stop condition If the source cannot be reproduced, label the conclusion as an observation or inference—not a verified finding.
Two different jobs share one name.
Australian providers currently advertise AI visibility audits, but the label alone does not tell you whether the work checks model answers, repairable public sources, or both. Compare the method before the score.
Prompt-presence snapshot
- Can show
- Record whether a named platform mentioned, cited, or described the business for exact questions under stated conditions.
- Cannot show
- Prove a stable recommendation rate, explain every model decision, or predict the answer another buyer will receive later.
- Useful when
- Useful when the method preserves prompts, conditions, repeated observations, citations, and limitations.
Public-source readiness review
- Can show
- Find crawl, fact, profile, markup, proof, and enquiry-path problems that weaken what customers and search systems can verify.
- Cannot show
- Guarantee that any AI system will include, cite, recommend, rank, or send demand to the business.
- Useful when
- Useful when every finding links to public evidence, an exact repair, an owner, and a done-when check.
Can the public business be verified six ways?
This on-device register checks repairable inputs: reachability, readable facts, Business Profile accuracy, cross-source agreement, truthful structured data, and public proof.
It deliberately does not test prompts or manufacture a visibility percentage. The result is a repair order for today's public evidence and nothing more.
Can the same business be verified six ways?
0 / 6 inspected
Can an ordinary visitor open the important service pages without a login, block, or broken route?
Are the real services, areas, hours, contact path, and proof stated in visible page text?
Is the profile verified, complete, and accurate for the business as it operates today?
Would a customer see the same name, service, area, hours, phone, and next step in both places?
Does LocalBusiness or Organization structured data describe only facts that a visitor can also verify?
Do recent genuine reviews, project details, and trusted public mentions support the service and area being claimed?
Verify the public inputs before drawing conclusions
Start with facts you can reproduce from the live website and Business Profile today.
All answers remain in this tab and disappear on refresh.
Six checks for the auditor—not the business.
A polished dashboard can still hide an unrepeatable method. Ask these questions before accepting a score, benchmark, or recommendation claim.
CONDITIONSDoes the report preserve platform, date, location, interface, signed-in state, model label, and exact prompts?
REPEATSDoes it show repeated observations or admit that a single answer is only a snapshot?
CITATIONSCan you open the cited sources and see how they support—or contradict—the answer?
FACTSAre name, services, area, hours, contact details, and proof checked against live public sources?
REPAIRSDoes every material finding name the change, owner, effort, and public done-when check?
BOUNDARYDoes the provider refuse guarantees and distinguish observation, inference, and uncertainty?
Run a twenty-minute check before buying a dashboard.
This is a small observation protocol, not a benchmark. Keep the record local, use no customer data, and stop if the process asks for credentials you do not need.
- 01
Declare the conditions
Write down the platform, interface, date, approximate location, signed-in state, model label when visible, and exact prompt.
- 02
Use real buying questions
Test a category-and-area question, an urgent-service question, and a comparison question a genuine customer might ask. Do not stuff the business name into a non-branded test.
- 03
Record the whole answer
Note which businesses appear, what is said, which sources are cited, and whether any detail is wrong. A screenshot alone is not a method.
- 04
Repeat before counting
Run the same small set again under documented conditions. Treat variation as a finding; do not hide it inside a single visibility score.
- 05
Trace back to public evidence
Open the cited pages and compare the live website, Business Profile, structured data, reviews, and third-party mentions.
- 06
End with repairs, not theatre
Prioritize corrections another person can reproduce and validate. Separate “not observed” from “cannot be found” and from “is incorrect.”
Know which audit Map & Mention sells.
The $129 Local Findability Audit is for one business and one location. It repairs the sources customers and search systems can verify; it does not pretend that six public checks are a cross-platform AI share-of-voice study.
- Included
- Public website and Business Profile evidence, reviews, structured facts, enquiry path, six-page PDF, exact repairs, validation checks, and a 30-day order.
- Not included
- Prompt tracking, competitor recommendation rates, private analytics, Search Console, account access, implementation, or platform guarantees.
- Best next action
- Run the register first. If the public facts are already wrong, repair them. If the evidence is scattered or priority is unclear, inspect the fictional report before buying.
Before you trust the score.
Five boundaries for a category where confident numbers can outrun the evidence.
What is an AI visibility audit?
The label can describe two different jobs: observing whether AI-assisted search interfaces mention or cite a business for documented questions, and checking whether the public sources those systems may use are crawlable, clear, consistent, and supported. A useful report states which job it performed.
Can an audit guarantee that ChatGPT, Gemini, Perplexity, or Google will recommend my business?
No. Answers vary with the system, model, time, location, account context, available sources, and the question itself. An audit can document observations and repair public evidence; it cannot promise future inclusion, citation, recommendation, ranking, traffic, enquiries, or revenue.
Can I run a basic AI visibility check myself?
Yes. Use a small set of real buying questions, preserve the exact conditions and answers, repeat the observations, inspect citations, and compare every business fact with the live public sources. Do not turn three prompts into a percentage benchmark.
Does the $129 Map & Mention audit include prompt tracking?
No. The founding Local Findability Audit is a focused public-evidence review for one business and one location. It checks the website, Business Profile, reviews, structured facts, and enquiry path, then delivers a six-page report and 30-day repair order. It does not sell prompt share-of-voice, competitor benchmarking, or a recommendation score.
What access does the Map & Mention audit require?
None. It reviews public evidence only. Do not send passwords, private analytics, customer lists, card details, bank details, or identity documents.
Repair the source trail before you pay to watch the score move.
Start with the six public inputs, inspect the fictional deliverable, and buy the smallest scope that removes a real decision.