What Should a GEO Checklist or Audit Template Include?

What does an AI visibility audit measure?

A complete audit answers six questions, in this order:

Layers 1 to 4 describe what is happening; 5 and 6 explain part of why. Off-site presence, meaning whether you appear on the lists, directories and publications engines cite, is usually the rest of the explanation and comes out of layer 3.

How do SEO, AEO and GEO audits differ?

The three overlap heavily, and the labels are used loosely, but the core difference is what gets measured:

A good GEO audit includes the SEO and AEO checks that matter for AI (crawlability, clear answers, structured data) but starts from observed answers. How the terms relate is covered in the difference between SEO, AEO and GEO.

How can you tell a GEO audit from a rebranded SEO audit?

Ask to see a sample report and check for these five things:

1. Real answers from AI engines, with the date, engine and exact question, not just "AI readiness" scores.
2. More than one run per question, with results shown as counts ("named in 3 of 5"), because single answers vary too much to judge.
3. Competitors named per question, not a generic competitor list.
4. Cited pages, showing which sources engines relied on.
5. Accuracy findings, meaning statements engines make about you that are wrong.

An audit that is mostly page speed, meta descriptions, backlinks and a few screenshots of ChatGPT is an SEO audit with a new label. Those checks are not useless, but they do not tell you whether AI engines recommend you or why not. See also whether low-cost GEO audit sites are reliable.

What goes into the audit template, section by section?

Use one document or workbook with these sections. Each item should record the evidence, not just a pass or fail.

1. Scope - Business, locations, product lines and languages in scope. - The question set: 20–30 buyer questions (category, requirement, comparison, problem, local), with exact wording and IDs. - Engines and modes tested, runs per question, test dates and locations.

2. Observed answers - One row per answer: date, engine, question ID, run, named (Y/N), position, businesses named, pages cited, description of you, evidence link. - Summary: appearance rate per question and engine, question coverage, share of voice.

3. Competitors - Businesses named at least three times, with the questions each one wins. - The question types each rival dominates, for example every comparison question.

4. Sources - Most-cited domains and pages across all answers, tagged by type. - Which cited third-party pages mention you, and which do not.

5. Accuracy - Every wrong or outdated statement, the engine that made it, and the likely source.

6. Crawler readiness and technical checks (details below).

7. Content coverage - Question-to-page map: best page per question, whether it answers directly, what is missing.

8. Entity and consistency - Business name, address, phone, category and core services compared across site, Google Business Profile, directories and social profiles. - Organization or LocalBusiness structured data matching the visible page.

9. Off-site presence - Presence on the directories, review sites, lists and publications engines cite for your category.

10. First-party AI data - Bing Webmaster Tools AI Performance (a sample of citations per page); Search Console's generative AI report (manual export; it also counts automated requests); AI-referred sessions and conversions in analytics, noting that GA4's AI Assistant channel leaves Perplexity and Claude in Referral; and, for merchants in the US, Canada, Australia, India and New Zealand, Merchant Center AI performance insights.

11. Findings and priorities - Each finding with evidence, affected questions, severity, effort and owner.

For collecting section 2 by hand, see how to manually track AI visibility.

What should an AI crawler readiness checklist include?

Crawler readiness is the gate: if engines cannot fetch and read a page, nothing else about it matters.

• robots.txt by purpose. List each AI crawler group you allow or block. OpenAI and Anthropic run separate crawlers for training, for their search index and for fetching pages when a user asks; Perplexity separates search from user fetches, and Google uses one crawler with a separate control token. Decide per purpose. Note that Google-Extended controls use for training Gemini models and for grounding answers in the Gemini app, and does not affect Google Search.
• Firewall, CDN and bot protection. Check that bot rules, rate limits and challenge pages do not block AI crawlers on key pages. Test with the crawlers' user agents and confirm in server logs.
• Status codes. Key pages return 200 to crawlers, not 403, 429 or challenge pages; redirects are single hops.
• Content without JavaScript. Main text, prices and facts are present in the raw HTML.
• Indexability directives. No accidental noindex, nosnippet or restrictive max-snippet on pages you want used.
• Discovery. XML sitemap is current; key pages are linked internally, not orphaned.
• Formats. Important facts are in HTML text, not only in images, PDFs or embedded widgets.
• Server logs. Recent visits from AI crawlers to key pages, as the final proof that access works.

Step-by-step commands for these checks are in how to audit your website for AI visibility gaps from its URL.

How is a GEO technical audit different from an SEO one?

Several technical assumptions from SEO do not hold for AI crawlers:

The practical result: a site that passes a standard SEO technical audit can still be unreadable to most AI crawlers if its key content depends on JavaScript, or blocked by a firewall rule nobody noticed.

How do you conduct an AI visibility audit step by step?

For one business with one location, allow roughly four to five working days:

1. Agree on scope and questions (half a day). Write and freeze the question set with the client or team, using sales and support wording.
2. Run the questions (one to two days). Several runs per question in each engine, in clean sessions, logging every answer. Automate where possible, but sample the real apps.
3. Analyse answers (half a day). Appearance rates, competitors per question, cited sources, accuracy errors.
4. Check the site (half a day). Crawler readiness, rendering, indexability, entity consistency.
5. Map coverage (half a day). Best page per question and the gaps.
6. Check off-site presence (a few hours). For the most-cited third-party pages, whether you appear and what they say.
7. Prioritise and write up (half a day). Findings with evidence, severity, effort and owners.

Keep all raw answers. They are the evidence behind every finding and the baseline for the next round.

What does an AI audit check for a small or local business?

The template scales down. For a single-location business, the same layers apply, but the weight shifts towards listings and reviews, because engines answering "best plumber near me" or "a dentist open on Saturday in Leeds" lean heavily on them:

• Questions with a place in them. Include the town, district and "near me" versions, plus service-specific questions ("emergency boiler repair", "invisible braces cost").
• Google Business Profile. Categories, services, hours, service area and description match what you actually offer, since Google's AI Overviews, AI Mode and Gemini draw on it heavily.
• Review sites and directories. Presence, rating and review count on the platforms that come up as cited sources for your trade and city.
• Service and location pages. One clear page per main service, with prices or price ranges where you can give them, the area you cover and what makes you different.
• Facts that go stale. Opening hours, prices, staff names and services you no longer offer, on your site and on every listing.

Most small-business audits end with a short fix list, often under ten items, and wrong or missing listing facts usually top it.

What should the audit report deliver?

A useful report is short at the top and complete underneath:

1. A one-page summary: current appearance rate across the question set, the three rivals named most, the most serious wrong facts, and the five most important fixes.
2. The findings table: every finding with evidence, affected questions, severity, effort and owner.
3. The question-level results: each question with results per engine, rivals named and sources cited.
4. The raw answer log, so anyone can check a finding against the actual answer.
5. The method: question set, engines, runs, dates and locations, so the next audit repeats it exactly.

Leave out any overall "AI score" that cannot be traced back to specific answers. If a number appears in the summary, a reader should be able to find the answers that produced it.

How should you prioritise audit findings?

Score each finding on severity and effort, then work through them in this order:

Within each group, start with the questions closest to a purchase.

How do agencies turn an audit into a GEO roadmap?

Agencies that audit for clients usually turn the prioritised findings into a 90-day plan:

• Days 1–30: fix access blockers and wrong facts; publish or rewrite the pages for the three to five most valuable missing questions.
• Days 31–60: work on off-site presence for the most-cited lists, directories and review sites; fix consistency across profiles.
• Days 61–90: remaining content gaps; re-run the full question set and compare with the baseline.

Each roadmap item should name the questions it is meant to change, so the re-run can show whether it worked. The baseline, the question set and the method must stay the same between rounds. For packaging this work, see what a GEO services package should include.

What do people ask about GEO audits?

How long does a GEO audit take?

For one business and one location, four to five working days by hand, most of it running and logging answers. Multiple locations, product lines or languages multiply the answer-collection part.

Can a GEO audit be fully automated?

The answer collection and technical checks can be largely automated. Judging accuracy, reading cited pages and deciding priorities still need a person who knows the business.

How often should a GEO audit be repeated?

A full audit every six to twelve months, with the question set re-run monthly or quarterly in between, is enough for most businesses. Re-run sooner after a redesign, migration or major content change.

What is the most common finding in GEO audits?

Missing pages for comparison and requirement questions, and absence from the third-party lists engines cite, are among the most frequent. Wrong facts copied from old listings or pages are the most urgent.