AI visibility audit checklist · 2026
Use these 20 checks to understand whether ChatGPT, Gemini, Perplexity and AI-assisted search can identify, verify and recommend your business — before you buy another dashboard or publish more generic content.
A useful AI visibility audit must test real buyer prompts, preserve the answers and citations, compare named competitors, verify what models believe about the brand and connect every finding to a technical, content or authority action.
What should an AI visibility audit actually answer?
The audit is not a search for one universal AI ranking. Generated answers change by model, prompt wording, user context, language and the sources available at that moment.
The goal is to establish a repeatable baseline: where the brand is mentioned, when it is recommended, which competitors occupy the shortlist, which pages or external domains are cited and whether the description of the business is accurate.
The checklist below separates five diagnostic layers. Keeping them separate prevents a rising mention count from being mistaken for qualified visibility.
Buyer-prompt baseline
- Define 15–30 questions based on real customer problems, comparisons and provider-selection decisions.
- Separate branded prompts from non-branded discovery; branded recognition is not the same as category visibility.
- Include market, language and geography where these change the commercial answer.
- Record the exact prompt, model, date and response rather than relying on memory or screenshots alone.
Mentions and recommendations
- Measure mention rate and recommendation rate separately; a passing mention may not influence the shortlist.
- Note how the brand is framed: category, audience, strengths, limitations and sentiment.
- Track which competitors repeatedly appear for the same decision context.
- Check whether product, service and location facts are current and accurate.
Citations and source influence
- Record every cited domain and page supporting the answer.
- Separate citations to your website from mentions supported only by third parties.
- Identify sources that repeatedly favour competitors: reviews, directories, media, communities or specialist pages.
- Check whether your strongest proof, service and comparison pages are accessible and internally linked.
Entity and technical readiness
- Confirm consistent brand name, offer, people, locations and contact information across key profiles.
- Validate canonical, language, indexation, crawlability, rendering and structured data on priority pages.
- Ensure important service and product facts appear in crawlable HTML, not only in images or scripts.
- Review whether About, author and organisation information connects the people and evidence behind the brand.
Content and commercial action
- Assign one primary intent to each service page, comparison, guide and case study.
- Answer the decisive buyer question early and support it with evidence, limitations and a next step.
- Map every visibility gap to an existing page, a genuinely distinct new asset or an external authority action.
- Prioritise by commercial relevance, evidence strength, effort and dependency — not by content volume.
How to score the baseline without inventing certainty
Use three separate scores instead of one opaque number. Each answers a different business question and can be tracked over time with the same prompt set.
Do not combine these into a promise of future rankings. A brand can be mentioned without recommendation, recommended without a link or cited while a competitor receives the commercial preference.
How to avoid keyword cannibalisation while acting on the audit
Keep the commercial page focused on the engagement: scope, process, deliverables, fit and CTA. Use informational articles to answer adjacent questions, explain measurement or teach the self-audit process.
- Commercial intent: AI Visibility Audit.
- Informational intent: AI visibility audit checklist.
- Ongoing service intent: AI Search and GEO optimisation.
- Proof intent: cases, experiments and observed citation changes.
When two drafts answer the same question, consolidate them. Internal links should identify which URL owns the commercial decision and which URL provides supporting education.
FAQ
Questions about auditing AI visibility
How often should the audit be repeated?
Use the same core prompt set monthly or quarterly depending on market velocity. Review the prompt set when products, audiences or buying behaviour change.
Can I run the checklist manually?
Yes. A spreadsheet and disciplined evidence capture can create a useful baseline. Dedicated tools help when prompt volume, model coverage or reporting frequency becomes difficult to maintain.
How many competitors should be included?
Choose the few competitors that genuinely appear in customer comparisons plus any unexpected brands repeatedly surfaced by AI answers.
Is AI traffic the same as AI visibility?
No. Mentions, citations, referral traffic and influenced conversions are different measurements and should not be collapsed into one metric.
Does llms.txt complete the technical audit?
No. It may communicate preferences to compatible systems, but it does not replace crawlability, indexation, useful pages, structured data or external evidence.
What happens after the audit?
Priorities usually fall into technical fixes, page improvements, new evidence assets, third-party authority and repeatable measurement. Only actions supported by the observed gaps should enter the roadmap.
Need the evidence, not just the checklist?
The fixed-scope audit maps buyer prompts, competitors, citations and site readiness into a practical 90-day roadmap.