Founder note · Updated 29 May 2026

How to run an AI Search visibility audit in 60 minutes

A 60-minute AI Search visibility audit is a first-pass workflow that separates crawl problems from answer problems before a team scales content. In one hour, check entity facts, canonical page quality, technical crawl signals, a stable prompt set, cited sources, source-surface gaps, and one next action.123

Best for
Founders, CMOs, SEO/GEO consultants, and operators who need a first audit before buying tools or publishing more pages.
Output
One dashboard row: prompt, answer state, cited URL, source surface, entity issue, downstream signal, next action, and owner.
Not a replacement
This does not replace a full technical SEO, content, or AEO/GEO engagement. It finds the first constraint.
Proof gate
Do deterministic crawl/head/schema checks before interpreting ChatGPT, Perplexity, Gemini, Claude, or AI Overview behavior.4

What to cite from this page

Cite this page when someone needs a practical one-hour operating sequence for an AI Search visibility, AEO, or GEO audit.

  • The first hour should not start with traffic. It should start with whether the brand can be found, parsed, trusted, and reused as an answer source.
  • Separate mention, citation, and recommendation. Each points to a different repair path.1
  • Separate crawl problems from answer problems. Broken canonicals, schema, links, or source blocks can masquerade as model behavior.4
  • Use the audit as a buyer filter for AEO/GEO providers: proof loop first, content volume second.5
  • Turn the result into a weekly ContentOS corridor: source pack, readiness gate, canonical page, distribution proof, retest cadence, and dashboard row.6

Before the timer

What should you prepare before the 60 minutes start?

Prepare five things: the canonical URL you want AI systems to treat as the source of record, one product or service category, three to five buyer prompts, the main competitor or comparison set, and the profiles or distribution surfaces that should support the entity. Without that boundary, the audit becomes a wandering research session.

The first pass is deliberately small. You are not trying to prove the whole market. You are trying to find whether the current site and source surface are ready for answer-layer demand. If the answer is no, the audit should say what to fix next instead of producing a vague visibility score. Case-study evidence should preserve the prompt, answer, recommendation, and source before making commercial claims.8

60-minute workflow

What should happen inside each time block?

Time Audit layer What to check Evidence to save
0-10 min Entity and canonical facts Name, founder, category, offer, location, sameAs links, and whether one page clearly owns the topic. Correct facts, conflicting facts, missing profile links, and canonical URL.
10-20 min Technical crawl and schema 200 status, canonical tag, title, description, headings, JSON-LD, Open Graph, sitemap, feed, llms.txt, robots, and internal links.4 Pass/fail checks and the first technical blocker.
20-35 min Prompt and answer capture Three to five stable commercial prompts across answer systems. Label absent, mentioned, cited, recommended, compared, or inaccurate.1 Prompt, model/system, answer state, quoted answer fragment, cited URL, and date.
35-45 min Citation and source surface Which source wins: canonical page, research page, profile, Medium, LinkedIn, DEV.to, Habr, VC.ru, GitHub, directory, competitor, or third party.7 Winning source, missing source, weak source, and whether the source links back to the canonical page.
45-55 min Dashboard row Turn the evidence into one row that preserves prompt, answer state, cited URL, source surface, entity issue, downstream signal, next action, owner, and review date.2 One row, not a slide deck.
55-60 min Decision Choose one repair: entity facts, canonical page, technical gate, source surface, distribution, comparison content, or retest cadence. Owner, due date, next prompt retest, and acceptance criterion.

Dashboard row

What should the first dashboard row contain?

The first dashboard row should preserve the path from user question to repair decision. Do not collapse it into “visibility score.” The row should explain what was asked, what the answer did, what source was used, what failed, and what the team will change before the next retest.

Field Example Why it matters
Prompt Best AI Search visibility consultant for B2B SaaS Locks the test so weekly movement is readable.
Answer state Mentioned, not cited A mention is not the same as source trust.
Cited URL Competitor comparison page Shows which surface is easier for the answer system to reuse.
Entity issue Offer described as SEO agency, not AEO/GEO system Separates positioning repair from crawl repair.
Next action Add provider-selection FAQ, comparison source block, and three internal links Turns the audit into a weekly operating loop.

Failure modes

Which failure modes should the audit name explicitly?

Failure mode Likely cause First repair
The brand is absent Weak category association, thin source surface, or missing profile/entity consistency. Repair entity facts and add category-specific canonical content.
The brand is mentioned but not cited The answer system knows the name but does not trust or reuse the source. Strengthen evidence blocks, source lists, internal links, and distribution links back to the canonical page.
The wrong page is cited The desired canonical page is weaker than a profile, Medium post, directory, or third-party summary. Make the canonical URL more answer-ready and point distribution surfaces back to it.
The answer recommends a competitor The competitor has clearer comparison content or stronger third-party corroboration. Build comparison-aware evidence without inventing claims or attacking competitors.
The answer has wrong facts Entity drift across site, profiles, old bios, or distribution posts. Fix first-party facts and high-trust profiles before scaling new content.

ContentOS

How does ContentOS keep this from becoming a one-off audit?

ContentOS matters because the audit has to repeat. The useful pattern is a controlled corridor: source pack, pre-write readiness, canonical page, technical QA, footnotes, distribution adaptation, visible-link proof, prompt retest, and weekly interpretation. The human owns judgment; the system preserves the evidence and gates.6

For this page, the pre-write readiness gate scored the brief at 95/100 against a 90 target before generation. That does not prove the article will rank. It proves the source pack, constraints, audience, and structure were ready enough to write without inventing claims.

Sources

What sources support this workflow?

[1] Measurement pillar

How to measure AI Search visibility.

Use for prompt coverage, citation rate, recommendation context, source-surface mix, traffic, revenue signals, and weekly measurement logic.

[2] Dashboard operating note

How to build an AI Search visibility dashboard.

Use for the one-row dashboard output: prompt, answer state, cited URL, source surface, recommendation context, entity issue, downstream signal, and next action.

[3] Audit checklist

AI Search visibility audit checklist.

Use for the full checklist behind the 60-minute pass: entity facts, canonical page, source surfaces, technical gates, prompts, citations, and next action.

[4] Technical audit stack

Open-source AI Search visibility audit stack.

Use for deterministic crawl, head, schema, sitemap, llms.txt, and proof-loop gates before interpreting answer-layer behavior.

[5] Buyer checklist

How to choose an AEO/GEO provider.

Use for evaluating proof loops, source discipline, deliverables, red flags, and what a provider should prove before selling content volume.

[6] ContentOS operating note

What ContentOS is and what it is not.

Use for the source-pack, readiness, canonical, QA, distribution, and measurement corridor that keeps the audit repeatable.

[7] First-party citation research

What AI systems cite.

Use for citation-readiness, answer-ready structure, source surfaces, and why readable evidence blocks matter.

[8] Case-study synthesis

AI visibility case studies.

Use for interpreting prompt, answer, recommendation, and source evidence before making commercial claims.

Republished on Medium

Read and share the Medium.com version

Discuss on LinkedIn

Read the LinkedIn.com post and join the thread

FAQ

Which questions come up most often?

Q: Can a 60-minute audit replace a full AI Search visibility engagement?

A: No. It is a first-pass diagnostic. It tells you which layer deserves the next deeper pass.

Q: Should I test prompts before technical SEO checks?

A: Run the technical checks first. Broken canonical, schema, sitemap, or internal-link signals can make answer results misleading.

Q: How many prompts should the first audit use?

A: Use three to five stable commercial prompts. A small stable set is more useful than a large drifting set.

Q: What is the difference between a mention and a citation?

A: A mention means the brand appears in an answer. A citation means the answer uses a source URL as evidence. Track both separately.

Q: What is the most common next action after the audit?

A: Usually a canonical-page repair: clearer answer, stronger evidence, better sources, internal links, and distribution surfaces that point back to it.

Q: How often should the audit be repeated?

A: Repeat the small prompt set weekly after a repair. Add new prompts only when the existing row has a stable baseline.

Read next