- Which buyer questions should an AEO audit test?
- What should the per-engine results show?
- The technical checks every AI search audit needs
- Why do third-party mentions belong in the audit?
- What you should receive from an AEO audit
- How do you judge the fix list and re-check?
- Where I fit if you want an AI search audit done for you
- Frequently asked questions
A useful AEO audit tests the questions your buyers ask across ChatGPT, Perplexity, Gemini and Google AI Overviews, shows who's named instead and which sources get cited, checks crawler access, entity clarity and page structure, and ends with a prioritised fix list and a re-check date. I'm Lilach Bullock, and after 21 years in marketing I now help small businesses see how AI assistants describe them, so I'm one option if you'd like an audit done for you.
Last checked: October 2026.
Commissioning an audit only makes sense if you know what you should get back. With AI search, plenty of reports look impressive and say very little: a colourful score, a few screenshots, a list of generic tips. This page is a buyer's checklist. Use it before you commission an audit to agree the scope, and afterwards to judge whether the report gives you something you can act on.
Some people call this a GEO audit, others an AI visibility audit or an AI search audit. The label matters less than the contents, so I'll stick to what should be inside.
Which buyer questions should an AEO audit test?
Everything starts with the question set. If the audit tests the wrong questions, every result after that is beside the point. The questions should match how your buyers ask, in their own words, not just your brand name or your favourite keyword.
- Questions people ask when choosing a provider, such as "who can help me with..." or "which tool should I use for...".
- Questions about the problem you solve, before people know your category exists.
- Comparison questions, where buyers weigh options against each other.
- Direct questions about your business, to see how assistants describe you.
You should see the full list in the report, and ideally agree it with the auditor before they start. If you can't see the questions, you can't judge the results.
What should the per-engine results show?
Each AI assistant behaves differently, so results should be broken down by engine: ChatGPT, Perplexity, Gemini, and Google AI Overviews or AI Mode. For each question you should be able to see whether your business was named, how it was described, which other businesses were named instead and which sources the assistant cited or linked.
That last part is often the most useful. The cited sources tell you where assistants are reading about your field. If the same few review sites, directories or articles keep coming up, those are the places worth being present on. The audit should also say whether each check ran with web search on or off, and note that answers vary from run to run.
The technical checks every AI search audit needs
Alongside the answers themselves, a solid audit looks at whether AI tools can read and understand your site at all. These checks are less exciting but they often explain why you're missing.
- Crawler access. Does your robots.txt allow the crawlers used by AI assistants, and do any security settings block them?
- Entity clarity. Is your business name, description and focus consistent across your site, social profiles and listings?
- Schema markup. Does your site use structured data for your organisation, people, services and FAQs, including sameAs links to your official profiles?
- About page. Does it say plainly what your business does and who you serve?
- Answer-first structure. Do your key pages answer the main question in the opening lines, or bury it under long introductions?
If you'd like to see how this fits into a longer plan, what a GEO strategy covers picks up where an audit leaves off.
Why do third-party mentions belong in the audit?
AI assistants don't only read your website. They read what others say about you: reviews, listings, interviews, articles, podcasts and directories. An audit that ignores this half of the picture will send you round in circles rewriting pages that were never the problem.
A good report lists where you're already mentioned, where your competitors are mentioned and you're not, and which of those sources assistants seem to trust. It should also flag outdated or wrong information about your business, because assistants can repeat it.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
What you should receive from an AEO audit
Here's a simple table to hold any audit against. If a section is missing, ask why.
| Audit section | What you should receive | Why it matters |
|---|---|---|
| Question set | The full list of buyer questions tested, agreed with you | Shows the audit reflects how your customers really ask |
| Per-engine results | Answers from each assistant, with web search noted | Each engine behaves differently, so one view misleads |
| Who's named instead | The businesses and sources named in your place | Shows who you're competing with inside AI answers |
| Crawler access | A check of robots.txt and any blocking | If AI tools can't read your site, nothing else helps |
| Entity clarity | Notes on naming, schema, sameAs and your about page | Helps assistants recognise your business without confusion |
| Page structure | Which key pages need answer-first rewrites | Clear answers are easier for assistants to quote |
| Third-party mentions | Where you're mentioned, missing or misdescribed | Assistants lean on what others say about you |
| Fix list | A prioritised list with the first steps marked | Turns findings into work you can start on |
| Re-check date | When the same questions will be tested again | Lets you see whether the changes made a difference |
How do you judge the fix list and re-check?
The fix list is where an audit earns its keep. It should be prioritised, not alphabetical, with the changes most likely to matter at the top and a short reason for each. It should say who'd do each fix: you, your web developer or whoever writes your content. Vague advice like "improve your content" isn't a fix. "Rewrite the opening of your services page so it answers who you help and how" is.
Finally, the report should end with a re-check date, when the same question set will be run again in the same way. Without it, you've got a snapshot and no way to know if the work helped. For the differences between this work and classic search, see where AEO and SEO overlap.
Where I fit if you want an AI search audit done for you
I run audits along the lines of this checklist for small businesses and marketing leads who want a clear, honest picture of how AI assistants see them, with fixes they can act on. With 21 years in marketing behind me, I care most about turning findings into plain next steps. I'm a good fit if you want one person to walk you through the results and help with the fixes. I'm not the right fit if you need ongoing reporting across dozens of markets or brands. You can read about my AI visibility consulting if that sounds useful.
Frequently asked questions
What is an AEO audit?
An AEO audit checks how AI assistants such as ChatGPT, Perplexity, Gemini and Google AI Overviews answer the questions your buyers ask, whether your business is named, and what on and off your site is helping or holding you back.
What's the difference between an AEO audit and an SEO audit?
An SEO audit focuses on how your pages rank in classic search results. An AEO audit looks at the answers AI assistants give, who they name and which sources they cite, and adds checks such as AI crawler access and entity clarity.
How often should I get an AEO audit?
Once to set a baseline, then a lighter re-check of the same questions every few months. AI answers shift, so regular re-checks matter more than repeating a full audit.
Can I do an AEO audit myself?
You can make a good start by asking ChatGPT, Perplexity and Gemini your buyers' questions and noting who gets named. A fuller audit adds technical checks, source analysis and a prioritised fix list.
What should I avoid in an AI search audit?
Be wary of reports built around a single score with no visible questions or answers, generic tips that could apply to anyone, and promises of guaranteed placement in AI answers.
