AI & TechnologyAI Search Visibility Audit: What It Checks and Costs

An AI search visibility audit checks two things: how often AI engines cite your site when buyers ask questions in your market, and whether the technical setup lets those engines find and read your pages. A fair audit measures citations over repeated runs, saves the evidence, and lists fixes in priority order. In 2026, published figures put one-time GEO audits anywhere from about $1,500 to $12,000, with a reported median of $4,300.

I run App369, a one-developer app and SEO/GEO studio in Miami, FL. App369 sells an AI search visibility audit, so weigh my opinions with that in mind. I have tried to keep this guide useful whether you buy from me, buy elsewhere, or do it yourself.

The two halves of a good audit

Most weak audits only do one half. They either run a few prompts and screenshot the result, or they run a technical crawl and call it "AI readiness." You need both, because one tells you where you stand and the other tells you why.

Half one: citation measurement

The audit starts with a list of real buyer questions. Not your brand name, but the questions a person types before they pick a vendor: "best invoicing software for small agencies," "how much does a website rebuild cost," "who can fix my Shopify page speed."

Each question gets run in the engines your buyers use: ChatGPT, Perplexity, Google AI Overviews and Gemini. Then the auditor records which sites each answer cites.

The key point is repetition. AI answers change between runs. The same question asked three times can cite three different sets of sources. So each question should be run several times per engine, and the result should be a share of runs: "cited in 4 of 9 runs." Never a rank. There is no stable position 1 to report, and anyone who says "you rank #3 in ChatGPT" is giving you a number the engine does not produce.

Good evidence also means saved answers: the full text, the cited links, and a screenshot, so you can check the work yourself.

Half two: technical checks

This half explains the citation results and finds fixes. It covers:

• Crawler access, bot by bot. Different bots do different jobs, and robots.txt treats them differently (details below).
• Indexing and snippet eligibility. Google says a page must be indexed and eligible to be shown in Search with a snippet to appear in AI Overviews or AI Mode. It also says there are no additional requirements or special optimizations.
• Structured data that matches the page. Google's guidance is to make sure structured data matches the visible text on the page. Markup that claims things the page does not say is a liability.
• Pages readable without JavaScript or a login. If the answer only appears after a script runs or a form is filled in, a bot may never see it. I check the raw HTML response.
• Sitemap and IndexNow. I check that the sitemap lists the pages you want found and that IndexNow, if you use it, is set up. Neither is a citation switch.
• llms.txt. Unproven. Google's AI features page says you do not need new machine-readable files or AI text files to appear in its AI features. I treat llms.txt as optional housekeeping, never the main fix. My separate guide on what llms.txt does goes deeper.

What each bot actually does

This is where audits most often go wrong, because people block or allow the wrong bot. These are the vendors' own descriptions, fetched on 2026-10-21.

Two takeaways. Blocking a training bot is a different decision from blocking a search bot. And Google-Extended is not a switch for AI Overviews: Google's page says it does not impact inclusion in Google Search or work as a ranking signal, and that page does not mention AI Overviews at all. Google's AI features page says AI Overviews follow the normal Search requirements.

For the step-by-step robots.txt setup, see how to get your SaaS cited in ChatGPT and Perplexity. For the bigger picture of how GEO fits with SEO, see the GEO guide.

Red flags when buying

• A guaranteed citation. Nobody controls what an engine says. A guarantee means the seller is guessing or selling.
• "Rank #1 in ChatGPT." There is no published ranking to hold. Ask what they mean.
• One run per question. A single run is an anecdote. Ask how many runs, and whether results are reported as a share.
• No saved answers or screenshots. If you cannot see the actual responses and cited links, you cannot check anything.
• llms.txt sold as the main fix. See above. It might be one line in a list of fixes, not the headline.
• No test conditions. Ask whether runs were logged out, from which location, and on what date. Personalized or signed-in answers do not generalize.
• A scope you cannot see. Ask for a sample report before you pay.

What it costs in 2026

I only cite figures I could open and read on the day I wrote this.

• First Page Sage (published October 7, 2026; quotes collected July 6 to October 2, 2026) reports a median one-time audit price of $4,300, with a typical range of $1,500 to $12,000 and a typical term of 2 to 6 weeks. Its median GEO/AEO retainer is $6,850 a month. The data covers 1,146 price quotes from 214 agencies. Two caveats from the report itself: it does not define the scope of an "audit," and only 23% of the agencies publish any GEO or AEO pricing. It is also one firm's own report on a market it competes in.
• Optimising (agency page, published June 30, 2025) described the market for one-off AI readiness audits as $10,000 to $20,000, "based on what we're seeing." It does not publish a price for its own audit.

The spread is wide because "audit" has no standard scope. The table shows how I would separate a cheap audit from a thorough one.

My own offer

App369's AI Search Visibility Audit is $750 fixed, delivered in 5 business days, and read-only (I do not change your site). I run 30 buyer questions, each 3 times per engine, in ChatGPT, Perplexity, Google AI Overviews and Gemini, logged out and from a US location. Every answer is saved with its cited links and a screenshot. Results are reported as citation share, with Gemini marked indicative. You also get technical findings and ranked fixes. If you then hire me for the $1,950 SEO + GEO Fix Sprint, the audit price is credited to it. More on the wider service is on the SEO, GEO and AEO page.

I will be plain about the limits. I am one developer, the audit measures a sample of questions at one moment, and it cannot promise any result. AI answers move.

Run a basic version yourself

You can learn a lot in an afternoon with no tools. Use a private window, signed out.

• Write 10 questions a buyer would ask before choosing a company like yours. No brand names.
• Run each question 3 times in ChatGPT, Perplexity, Google (look for an AI Overview) and Gemini.
• For every run, save the answer text, the cited links and a screenshot. Note the date and your location.
• Count runs where your domain is cited, and divide by total runs. Do the same for two competitors.
• Open your robots.txt. Check for rules on OAI-SearchBot, GPTBot, PerplexityBot, ClaudeBot and Google-Extended, and confirm each rule says what you intend.
• Confirm your key pages are indexed in Google Search Console and not marked noindex or blocked by snippet limits.
• View the page source (not the rendered page) of your key pages. Is the main answer in the HTML?
• Check that no key page needs a login.
• Compare your structured data with the visible text. Remove claims the page does not make.
• Confirm your sitemap lists the pages you want found and is submitted in Search Console.
• Write down the top three gaps and fix the cheapest one first.

If the results are confusing, or you would rather have it done with saved evidence, that is when a paid audit makes sense.