Skip to content
AI SEO

How to Run an SEO Audit With AI: A Checklist for Google and AI Search

November 11, 20266 min readAdvik Jain

Co-founder & CEO, Optivus Technologies

In short

An AI SEO audit checks two things: whether Google can crawl, index and trust your pages, and whether AI search engines can reach and quote them. Use a crawler and Search Console for the data, then use AI to explain issues, group them by cause and rank fixes by impact. Add the AI-specific checks: robots.txt rules for OAI-SearchBot and PerplexityBot, answer-first content, sourced claims and consistent facts about your business. Fix the issues that block crawling and indexing first; everything else depends on them.

Key takeaways

  • An AI SEO audit covers two readers: Google's crawler and the AI engines that decide what to cite. One crawl can check both.
  • AI speeds up the analysis, not the data collection. Crawlers and Search Console find the issues; AI explains them, groups them by cause and ranks fixes.
  • Core Web Vitals have published targets: LCP within 2.5 seconds, INP of 200 milliseconds or less and CLS of 0.1 or less, measured at the 75th percentile of page loads.
  • Check robots.txt by bot. Blocking OAI-SearchBot keeps you out of ChatGPT search answers, while Google-Extended controls Gemini training and does not affect Google Search.
  • Fix in order: crawling and indexing, then duplicates and speed, then content quality and answer-readiness.

What is an AI SEO audit?

An AI SEO audit checks your site for two readers. The first is Google's crawler: can it crawl, index and trust your pages? The second is AI search: can ChatGPT, Perplexity, Gemini and Google's AI Overviews reach your pages and find answers worth quoting? The two overlap almost entirely, because Google's AI features draw on the same index as its search results, but AI search adds a few checks of its own.

"AI" also describes how you run the audit. Crawlers and Search Console collect the data; AI then explains each issue, groups issues by cause and ranks fixes by likely impact. That turns a spreadsheet of 3,000 warnings into a short list of things to do.

The checklist

AreaWhat to checkData from
Crawlingrobots.txt, server errors, redirect chains, blocked resourcesCrawler, Search Console
IndexingPages excluded from the index and why; sitemaps; noindex tagsSearch Console Page indexing report
DuplicatesCanonical tags, near-duplicate pages, parameter URLsCrawler
SpeedCore Web Vitals: LCP, INP, CLSPageSpeed Insights, Search Console
On-pageTitles, meta descriptions, one H1, heading orderCrawler
Internal linksOrphan pages, broken links, important pages buried deepCrawler
Structured dataValid markup that matches visible contentSchema validator
Content qualityThin, outdated or duplicate pages; unsourced claimsCrawler plus a person
AI crawler accessRules for OAI-SearchBot, PerplexityBot, GPTBot, Google-Extendedrobots.txt, CDN or firewall settings
Answer-readinessDirect answers under question-led headings; tables for comparisonsA person, with AI help
Brand factsPrices, locations and product names consistent everywhereYour site, profiles, AI answers

The thresholds that matter

Google's Core Web Vitals targets for a good page experience are:

  • Largest Contentful Paint (LCP): within 2.5 seconds of the page starting to load.
  • Interaction to Next Paint (INP): 200 milliseconds or less.
  • Cumulative Layout Shift (CLS): 0.1 or less.

Each is measured at the 75th percentile of page loads, separately for mobile and desktop. A page that meets them for most visits but not the slowest quarter still fails.

The AI search checks

Crawler access, bot by bot

Different bots do different jobs, and robots.txt treats them separately:

  • OAI-SearchBot, which OpenAI's crawler documentation says surfaces websites in ChatGPT's search features. Sites that opt out "will not be shown in ChatGPT search answers."
  • PerplexityBot, which Perplexity's documentation says is "designed to surface and link websites in search results on Perplexity."
  • GPTBot, which collects content that may be used to train OpenAI's models.
  • Google-Extended, which Google's crawler documentation says controls whether content may be used for training future Gemini models, and "does not impact a site's inclusion in Google Search."

Also check your CDN or firewall: bot protection can block AI crawlers even when robots.txt allows them. The AI crawler audit covers sixteen bots in more depth, and a free AI crawler checker tests your site.

Answer-readiness

For your ten most important pages, check that each one states its main answer in the first two sentences under a heading phrased as a question, uses a table where readers compare options, and has sections that make sense on their own. AI answers lift passages, not whole pages.

Sourced claims and consistent facts

AI engines repeat what they read. Every statistic on your site should link to its source, and every fact about your business (prices, plans, locations, product names) should match across your site and your profiles elsewhere. How to audit content for unsourced claims is a step-by-step method.

How to run it with AI

  1. Crawl and export. Run a full crawl and export the Page indexing and Performance reports from Search Console.
  2. Group by cause. Ask AI to group issues by root cause and page template. A missing canonical on a template is one fix, not 900 warnings.
  3. Rank by impact. Ask it to order groups by how many important pages each affects and whether it blocks crawling or indexing.
  4. Explain and draft fixes. Have AI explain each issue in plain language and draft the change, such as a robots.txt rule, a redirect map or a schema block.
  5. Check before shipping. A person confirms each fix suits your setup; test code on a staging site.
  6. Re-crawl and compare. Confirm the fixes landed and nothing new broke.

A prompt that works for step 2:

Here is a crawl export of issues with URLs and page types: [paste]. Group the issues by root cause, count affected pages, and say which group blocks crawling or indexing. Do not invent issues that are not in the data.

Fix order

PriorityWhatWhy first
1Crawling and indexing blockersNothing else matters for a page search cannot index
2AI crawler access for search botsRemoves you from AI answers if wrong
3Duplicates and canonicalsSplits signals between pages
4Core Web Vitals on key templatesAffects every page using the template
5Content quality and answer-readinessDecides whether indexed pages rank and get quoted
6On-page polishTitles, meta descriptions, headings

Self-check

  1. Are all your important pages indexed?
  2. Do your key templates meet the Core Web Vitals targets at the 75th percentile?
  3. Does robots.txt, and your CDN, let the AI search bots you want reach your site?
  4. Does each key page answer its main question in the first two sentences?
  5. Do the facts about your business match everywhere they appear?

Risks and how to manage them

RiskHow to manage it
Thousands of warnings, no actionGroup by cause and template; fix by impact
AI explaining issues that are not thereGive it the export; forbid invented issues; spot-check
A fix that breaks something elseTest on staging; re-crawl after shipping
Blocking AI search bots by accidentReview robots.txt and CDN bot rules after every change
Auditing once and forgettingAutomated crawls every week or two; a quarterly deep review

Where Veritas fits

Veritas runs this audit on a schedule. Its site audit checks pages for Google's crawler and AI answer engines in one crawl, every two weeks, and groups every fix by severity. AI visibility tracking adds the other half, what ChatGPT, Perplexity, Gemini and Google AI Overviews actually say about you, and the free tools cover one-off checks such as robots.txt, indexability and schema.

The bottom line

An AI SEO audit is a regular SEO audit plus the checks AI search adds: bot-by-bot crawler access, answer-ready pages, sourced claims and consistent facts. Let crawlers and Search Console collect the data, let AI group and explain it, and fix crawling and indexing before anything else.


Want an audit that checks for Google and AI crawlers in one crawl? Get started with Veritas.

Related reading: The AI Crawler Audit: Sixteen Bots and What Each Does With Your Pages · How to Use AI for SEO: A Step-by-Step Workflow.

Want to see which AI answers mention your brand?Start a 7-day trialSee pricing

Frequently asked questions

Can AI do an SEO audit by itself?

It can analyze one, not collect the data. A language model cannot crawl your whole site or see your Search Console data unless a tool supplies it. Pair a crawler and Search Console with AI for explanations and priorities, and have a person decide what to fix.

How often should I audit my site?

Run an automated crawl every week or two and a deeper review each quarter, plus a check after any redesign, migration or major plugin change. Those are the moments when crawling and indexing break.

What is the difference between an SEO audit and an AI visibility audit?

An SEO audit checks whether search engines can crawl, index and rank your pages. An AI visibility audit checks what AI engines say about you and which pages they cite. An AI SEO audit covers both, because AI answers depend on the first.

Should I block AI crawlers in robots.txt?

Decide bot by bot. Blocking search crawlers such as OAI-SearchBot or PerplexityBot removes you from those engines' answers; blocking training crawlers such as GPTBot or Google-Extended limits model training without affecting Google Search.

How much does an AI SEO audit cost?

A self-run audit costs the tools you already use. Agencies often price a one-off audit as a project; in an Ahrefs poll of SEO providers, $2,501 to $5,000 was the most common per-project fee. Ask what the audit includes before comparing prices.

Which free tools can I use for an AI SEO audit?

Google Search Console for indexing and performance, PageSpeed Insights for Core Web Vitals, and free checkers for robots.txt, AI crawler access, indexability and schema. Veritas lists several free checkers on its tools page.

Build content that gets cited.

Veritas drafts marketing content from your own documents with each knowledge-based claim cited to its source, and tracks whether AI answers mention you.