An AI visibility audit can cost nothing, and its price tells you less than what the audit actually samples. At Probably Genius, the 109-point Recommendation Check is free. For any paid audit, the fee should track how many buyer questions it asks, on which engines, how many times, and how much human scoring sits behind the result.
If a proposal labeled "AI visibility audit" has landed beside your SEO retainer, it is fair to wonder whether it sells a diagnosis, a product or the whole job. That confusion isn't a judgment on you. Nobody has drawn the line for you yet.
Think of a home inspection report. It tells you where the damp is and what the repair will roughly involve. It doesn't fix the roof. Below: what an audit is for, what its fee pays for, the two layers it should check, and why the result goes stale.
What an AI Visibility Audit Is For
An AI visibility audit measures one thing: whether AI assistants can find your business, describe it accurately and name it when buyers ask questions in your category. A good one also ranks the gaps, so you know which ones matter first.
Google's guidance on hiring search help puts the job description in one sentence. An SEO audit, it says, "should be about giving you realistic estimates of improvement, and an estimate of the work involved." That standard transfers neatly to AI visibility. The audit is a diagnosis plus a work estimate.
So the useful question to ask any provider is simple: what does your audit end with? If the answer is a score and a sales call, you bought a lead magnet. If the answer is a ranked list of gaps, the evidence behind each one, and a plain estimate of the work to close them, you bought an audit.
Where the audit stops matters as much as where it starts. It hands you a list. It doesn't write the pages, repair the facts or earn the third-party mentions on that list. Those are separate purchases. We mapped all three price bands (software, add-on work and an operated program) in how much AI visibility optimization costs.
What a Fair Audit Fee Is Paying For
An audit's cost is mostly the cost of its sample and its scoring. Three numbers decide the sample: how many buyer questions are asked, how many engines answer them, and how many times each question is run.
Ask for the repeat count in writing. It matters most. SparkToro's January 2026 study had 600 volunteers run 12 prompts through ChatGPT, Claude and Google's AI a combined 2,961 times. For ChatGPT and Google's AI, the chance of getting the same list of brands in two responses was under 1 in 100; Claude was only slightly more likely to repeat itself. The study was run with an AI tracking vendor and is not peer reviewed, but the direction is hard to argue with. Answers wobble.
One run per question is one roll of the dice. An audit built on single screenshots can tell you a story. It can't tell you a rate.
The second cost is scoring. Someone has to read the answers, check what each engine said about you against what is true, and inspect the site against a fixed checklist. Ours is 109 specific checkpoints across eight zones, hand-scored. That reading takes people, and a person's hours cost more than a scheduled scan.
The two costs also explain why prices for things called an audit can sit so far apart. A scan that asks one engine ten questions once, and grades your homepage automatically, costs almost nothing to run. A reading that asks forty questions across five engines several times each, keeps every answer and has a person check each description against the truth costs real hours. Both are sold under the same word. Only the second gives you a rate you can act on, and only the second can be re-run later as a fair comparison.
You'll notice there's no market rate card in this section. That's deliberate. We looked for a methodical survey of audit fees and found vendor claims without a method behind them. So instead of a band nobody can check, here is the test: judge any fee against its sample, its scoring and what it ends with.
Before you pay, ask for four things in writing:
- The number of buyer questions, the engines, and the runs per question.
- Whether the raw answers are saved, so you can compare them later.
- Who scores the results, and against which checklist.
- An estimate of the work, at the end of the report.
Layer One: Can the Machines Reach You at All?
An honest audit checks access before it checks reputation. If the engines can't reach your pages, nothing written on them counts.
Google's documentation for its AI features is clear about the bar. To be shown as a supporting link in AI Overviews or AI Mode, "a page must be indexed and eligible to be shown in Google Search with a snippet." Its list of fundamentals starts with "ensuring that crawling is allowed in robots.txt, and by any CDN or hosting infrastructure." It also asks that structured data match the visible text and that your Business Profile be current.
OpenAI's crawler documentation adds a detail many owners have never heard. Sites that opt out of its search crawler, OAI-SearchBot, "will not be shown in ChatGPT search answers, though can still appear as navigational links." A security setting at your hosting provider can do that opting out for you, silently. After a fix, OpenAI says robots.txt changes take about 24 hours to reach its search systems.
This is also where a familiar red flag shows up. Google says plainly that you do not need "new machine readable files, AI text files, or markup" to appear in its AI features. An audit whose main recommendation is a special AI file has sold you packaging. Skip it.
The good news for anyone already paying an SEO agency: this layer is where a technical site audit usually overlaps, and that overlap is fine. Your agency may already have it covered. Ask them to confirm the crawler settings by name.
Layer Two: What the Engines Say When Buyers Ask
The second layer is the one a technical SEO audit rarely runs. It asks the engines your buyers' real questions, and records what comes back.
We test the five engines your buyers actually ask: ChatGPT, Gemini, Perplexity, Claude and Grok, with the questions they actually ask. Each answer gets read for three things: whether you were named, whether you were described accurately, and who was named instead.
Report it as a rate, never a position. Given how rarely the same list repeats, "you rank third in ChatGPT" describes one roll, not a standing. "You appeared in four of twenty answers, accurately in three" describes a pattern you can work on.
The fix list deserves the same scrutiny as the score. A controlled study posted in May 2026 ran 252,000 trials across six language models and found topical relevance and list position were the biggest drivers of being cited first. Explicit price information and a recent timestamp helped consistently, while formatting-only edits had little impact. It was a testbed rather than a live assistant, so read it as direction. The direction still says the expensive part of the fix is relevant, specific material, and a fix list full of formatting tweaks is light on substance.
Why an Audit Has a Date on It
The answers an audit records do not hold still. SISTRIX tracked 82,619 prompts for 17 weeks, from December 2025 to April 2026, across six countries including the United States. Google's AI Mode replaced 56 percent of its cited sources every week. ChatGPT Search replaced 74 percent.
AI Overviews were steadier for some questions: in 53 percent of prompts, not a single source changed over the 17 weeks. In 28 percent there were isolated changes, and in the remaining 19 percent sources drifted as much as in AI Mode. SISTRIX's own conclusion was blunt: at that rate, a single citation placement "is not a reproducible result but merely a snapshot."
That changes what you are buying. A one-time audit is a dated reading of a moving system. It's genuinely useful. It also has a shelf life.
So put a re-run date on any audit before you pay for it, and insist the re-run asks the same questions on the same engines. Only like-for-like readings can tell you whether anything moved. Measurement shows the gap. Operations close it, and a second reading shows whether they did.
From Audit to Program, Without a Second Invoice Surprise
Here's how the sequence runs with us, stated plainly so you can compare it with anyone else's.
The Recommendation Check comes first. It is free, presented in about an hour, with no obligation, and the findings are yours either way. It reads your market: who AI recommends instead of you, where you are invisible, and what the gaps would take to close.
If you decide to do something about it, the plan follows how contested your market is, not how much revenue your business makes. The Answer, our plan for competitive, winnable markets, is $2,500 setup + $1,500/month, twelve-month program. It runs 12 done-for-you thought-leadership articles a month, planned to create approximately 60 clear Knowledge Entries, alongside press, entity work and monthly five-engine monitoring. Smaller markets get smaller plans, and the full ladder sits on our pricing page.
Your SEO agency can stay. It works best when everyone knows who owns what, which is exactly what a good audit makes visible. Bring the reading to your next agency review and walk through the gaps together. Usually some are covered, some are measured and nobody owns the rest. That last group is the point of the exercise. Nobody gets fired to find it.
The method behind the check is public, and you can read every part of it in our methodologies. You already have the genius. We make sure AI can see it.