What should an AI visibility audit include?

It should document the market, prompts, systems, dates, brand mentions, cited sources, factual accuracy, competitor appearances, technical access evidence, limitations and remediation priorities.

Can I print this AI visibility checklist?

Yes. The checklist is formatted for browser use and printing.

Does completing the checklist guarantee AI citations?

No. The checklist supports evidence gathering and diagnosis; no audit can guarantee selection or citation by an AI system.

AI Visibility Audit Checklist

How to use it: define one market and one audience, complete every applicable item, and retain screenshots or exports with dates. Tick boxes stay in your current browser session only; print or save a PDF if you need a retained working copy.

01 · Scope

Define the commercial question

Market and language recordedCountry, language and any meaningful regional context are fixed before testing.

Audience and buyer role definedThe benchmark names the customer type, decision-maker and purchase stage.

Category and offer boundaries documentedThe products, services and exclusions being measured are unambiguous.

Known competitors listed separatelyDo not force them into every prompt; preserve them as a comparison set.

Market / audience

Category / offer

02 · Prompts

Freeze a representative prompt set

Category-discovery prompts includedQuestions that ask which providers or solutions exist.

Problem-aware prompts includedQuestions framed around the customer’s problem before a provider is known.

Comparison and trust prompts includedQuestions involving alternatives, evidence, risk, specialisation or suitability.

Branded accuracy prompts includedQuestions checking whether the organisation and offer are described correctly.

Exact wording and prompt IDs frozenCore wording must remain stable across engines and reporting periods.

Core prompt count

Prompt-set version

03 · Answer evidence

Capture results consistently

System, product mode, date and market captured“ChatGPT” or “Google” alone may be insufficient to reproduce the context.

Brand presence and prominence recordedRecord whether the brand appears, where it appears and how it is framed.

Cited domains and URLs retainedKeep the sources, not merely the generated answer.

Accuracy checked claim by claimFlag incorrect services, locations, ownership, prices or unsupported assertions.

Competitor appearances recorded neutrallyCapture repeated alternatives without assuming every mention is commercially positive.

Raw evidence retainedScreenshots, exports or records are labelled with prompt ID and timestamp.

Systems tested

Evidence location

04 · Technical access

Test retrieval as a separate layer

Canonical origin and tested paths recordedSubdomains, redirects and alternate origins are not silently combined.

robots.txt captured and interpreted by identitySearch, user-fetch, training and data-use identities are not treated as one generic bot.

Sitemaps and optional machine guidance observedAbsence of an optional llms.txt file is not automatically marked as a failure.

Declared policy separated from observed accessA permitted policy does not prove that a CDN or WAF delivered clean content.

Challenges, denials, redirects and timeouts classifiedUnexpected HTML and challenge pages are not mistaken for valid robots files or content.

Origin / paths

Technical evidence reference

05 · Decisions

Turn findings into bounded actions

Confirmed findings separated from inferenceEvery conclusion points to evidence or is labelled as an assessment.

Intentional policy choices preservedTraining restrictions are not removed simply because retrieval access is desirable.

Remediation prioritised by impact and certaintyDo the confirmed, high-impact work before speculative optimisation.

Owner and verification step assignedEach change has a responsible person and a method for confirming the result.

Baseline retained for the next reviewCore prompts, conditions and technical evidence remain available for comparison.

Top priority / owner

Re-test date

Technical access decision gate

No confirmed barrier

Keep the evidence. Move separately to entity, authority, content and answer-selection analysis.

Could not assess

Do not guess. Resolve the assessment limit or commission deeper infrastructure investigation.

Confirmed barrier

Prioritise the exact policy or delivery fix, then re-test before attributing later visibility changes.

This checklist cannot prove future AI selection. Answer outputs vary by system, market, prompt, account context and time. Technical access is a prerequisite, not a citation or ranking guarantee.

Need the technical evidence completed for you?

Digital Dominator’s fixed-price Australian audits are human-reviewed, delivered within five business days and include developer-ready remediation. Prices include GST.

Access Check · A$495 Full Audit · A$1,195