The two audits sound like the same product with a newer label. They are not. An SEO audit looks at your website and asks whether Google can crawl it, index it, and rank it. An AI visibility audit looks at the assistants' answers and asks whether you were in them. One starts from your pages. The other starts from the buyer's question. This article sets out what each one actually finds, the two places they overlap, and a rule for which to run first that does not depend on which one you are being sold.
What an SEO audit finds
An SEO audit is an inspection of the site. A good one covers whether the crawler can reach every page that matters, whether those pages are indexed, how fast they load, whether the titles and headings say what the page is about, whether internal links carry weight to the pages that need it, and where the site ranks for the terms it targets. The output is a list of defects in your site and the fixes for each, usually ordered by how much ranking they cost you.
Its blind spot is the answer. An SEO audit can tell you that your pricing page ranks fourth for "workforce management software pricing". It cannot tell you that when a buyer asks ChatGPT the same question, the answer names three competitors and never you, because it never asked ChatGPT anything. It measures your standing in a results list. The buyer increasingly never sees that list.
What an AI visibility audit finds
An AI visibility audit is an inspection of the answers. It takes the questions your buyers ask, puts each one to the assistants in real sessions, and records what came back: who was named, who was linked, what was said about each, and which sites the assistant leaned on. The output is a picture of the answer your buyer sees, with the recorded text behind it, and a list of the sites taking the answers you should have.
Three things separate a sound one from a screenshot. It runs each question more than once, because the same question gets different answers on different runs and a single run can flatter or damn you by chance. It runs on fresh sessions with memory off, so the assistant's picture of the person asking does not leak into the result. And it shows the recorded answers, not just a score, so a claim in the report can be checked against the sentence it came from.
Its blind spot is your site. It can tell you that Perplexity cited a competitor's comparison page for a question you should own. It does not, on its own, tell you that your equivalent page takes six seconds to load and is blocked to two of the crawlers. That is the SEO audit's territory, and the reason the two are usually run together.
The SEO audit asks whether Google can find you. The AI visibility audit asks whether the assistants choose you once they have.
Where they overlap
Two findings show up in both, and both are worth checking twice.
Crawler access. The assistants fetch pages with their own crawlers, and a firewall rule or a robots directive that keeps them out is a defect in your site with a symptom in the answers. An SEO audit will find the rule. An AI visibility audit will find the symptom: you are never cited, even for your own name. Only the second tells you what it is costing you; only the first tells you where to fix it.
The search results the assistants read. Most assistants answer by running searches of their own and reading what comes back. A page that does not appear in those results cannot be cited, whatever else is true of it. So the rankings an SEO audit reports are an input to what the AI visibility audit measures, and a page that ranks well is still the most reliable way to get cited. The overlap is not an accident. The assistants are built on top of search.
Which one you need
The rule is short. Ask whether your site already ranks for the terms your buyers use.
- If it does, run the AI visibility audit. Your open question is whether the assistants pick you when they have the chance, and an SEO audit has no way to see that. The findings will be about the pages you are missing and the sites winning them, not about your site's plumbing.
- If it does not, run the SEO audit first. A site the assistants' searches never return cannot be cited, and an AI visibility audit will spend its pages telling you that in ten different ways. Fix the ranking problem, then measure the answers.
- If you are not sure, the AI visibility audit will tell you, because retrieved and cited are recorded separately. Never retrieved is a search problem. Retrieved but never cited is a page problem. The two have different fixes.
Most established businesses end up with both, in that order: a measurement of the answers, then the technical work the measurement showed to be the bottleneck. Our free audit bundles the crawler check and the site's speed and structure into the measurement, which is why it reads as both, but the ordering above still holds when you decide what to fix first.
Three questions to ask whoever is selling you either one
- How many times did you run each question? Once is a screenshot. The answer should be a number, and it should be more than one.
- Can I see the recorded answers? If the report is a score and a chart with no text behind it, there is nothing to check.
- What do you do with the finding? A finding without a page to build or a rule to change is a diagnosis without a prescription. The report should name the fix.
The full list of questions, with what a good answer sounds like, is in questions to ask an AI SEO agency. The protocol we use for every audit is on the methodology page, and the instrument that runs it is Citrion.
Find out which problem you have
The free audit records what the assistants answer for your buyers' questions, with the crawler check and the site's speed included. It tells you whether your bottleneck is search or the page.
Get my free auditSources
- Google Search Central, "AI features and your website" (how AI Overviews and AI Mode draw on search results and standard crawling): developers.google.com
- OpenAI, "Overview of OpenAI crawlers" (OAI-SearchBot and how ChatGPT search fetches pages): platform.openai.com
- EZ Web, "How we measure AI visibility" (the protocol behind every audit: fresh sessions, memory off, multiple runs, recorded answers): ezweb.dev/methodology