A buyer opens ChatGPT and types: "best project management tool for a small construction company." The assistant names four products, explains why each fits, and the buyer shortlists two of them. Your product would have been perfect. It wasn't mentioned. You'll never know the deal existed.
This is the new silent loss. Buyers increasingly ask AI assistants, ChatGPT, Gemini, Claude, Perplexity, for recommendations the way they once asked Google, except the assistant doesn't return ten blue links. It returns an answer, with names in it. If your brand isn't one of those names, you're not on page two. You're nowhere, and no analytics dashboard will tell you.
An AI visibility audit is how you find out where you stand in these answers, what's shaping them, and what to do about the gaps. Here's how to run one this week.
What an AI visibility audit is
An AI visibility audit is a structured check of how AI assistants represent your brand across the questions your buyers actually ask. It measures three things:
- Presence: do assistants mention you at all when buyers ask for recommendations in your category?
- Accuracy: when assistants describe you, do they get your positioning, pricing, and audience right, or are they working from a three-year-old version of your website?
- Framing: who do assistants compare you to, what do they say you're best for, and what caveats do they attach?
Think of it as a brand perception audit for a new audience, except the audience is a handful of language models that millions of buyers now treat as a trusted advisor.
Why this matters more every quarter
Search behavior is shifting from queries to questions, and from links to answers. When someone asks an assistant "what CRM should a 10-person agency use," the recommendation step, the part where your SEO, your ads, and your homepage used to compete for attention, happens inside the model's answer. The buyer may visit only the two or three sites the assistant named.
Three properties make this shift uncomfortable for brands:
- It's winner-take-most. An answer has room for three to five names. There's no page two to limp onto.
- It's invisible. You can't see the conversations you were absent from. The losses never register anywhere.
- It compounds. Assistants synthesize what's written about you across the web. Weak presence today shapes the content of tomorrow's answers, and the gap widens quietly.
What actually shapes AI answers
You can't buy your way into an assistant's recommendation, and you can't keyword-stuff your way in either. Models build their picture of your brand from the corpus of text about you, so the levers are about clarity and corroboration.
Entity clarity
Does the model understand what you are? Brands with a clear, consistent description, what category they're in, who they serve, what they do, get represented accurately. Brands with vague taglines and shifting self-descriptions confuse models the same way they confuse buyers. If your own website can't say plainly what you are, an assistant certainly can't.
Consistent positioning across sources
Models cross-reference. If your website, your social profiles, your directory listings, and third-party articles all describe you the same way, that description hardens into the model's answer. If they conflict, the model hedges, generalizes, or skips you for a brand it's surer about.
Reviews and third-party validation
Assistants lean heavily on what others say about you: review platforms, comparison articles, forum threads. A brand with substantial, specific review coverage gives the model both confidence to recommend it and language to describe it. This is the same dynamic we covered in how customer reviews reveal your real positioning, with a twist: now the machines are reading your reviews too, and repeating what they find.
Structured, crawlable information
Clear pricing pages, feature descriptions in plain language, FAQ content that answers real buyer questions, schema markup. Anything that makes facts about you easy to extract makes you easier to include in an answer. Models recommend what they can describe.
How to check your AI visibility today
You can run a first-pass audit manually in an afternoon. Open ChatGPT, Gemini, and at least one other assistant, and run these prompt types. Use fresh sessions so your history doesn't contaminate results.
- Category prompts: "best [your category] for [your ideal customer]". Run five variations with different customer types and use cases. Note every brand named, and whether you appear.
- Direct prompts: "what is [your brand]?" and "is [your brand] good for [use case]?". Check the description against reality. Stale facts and wrong audiences are findings.
- Comparison prompts: "[your brand] vs [competitor]" and "alternatives to [competitor]". See how you're framed, and whether you appear in your rivals' alternative lists.
- Problem prompts: describe the problem you solve without naming any product. "How do I [the job your buyers hire you for]?" Do you come up as part of the solution?
Log everything in a spreadsheet: prompt, assistant, brands named, how you were described, errors. Repeat monthly, because answers shift as models and their sources update. And run your top competitors through the same prompts while you're at it; knowing they're absent too changes the finding from a crisis into a land grab.
Fixing the gaps you find
The findings tend to sort into three buckets, each with a different fix.
If you're absent from category prompts, the problem is usually entity clarity and third-party corroboration. Tighten your category language so every page says what you are in plain words, get listed in the directories and comparison articles that cover your space, and build review volume on the platforms assistants cite. You're trying to give models more, and more consistent, raw material.
If you're described inaccurately, audit your own properties first. Old positioning lingering on about pages, outdated pricing in cached content, conflicting descriptions between your site and your profiles. Models repeat the contradictions they find. Make the current truth the loudest version everywhere.
If you're framed badly in comparisons, look at what comparison content exists. If the only "you vs competitor" articles on the web were written by your competitor, the model learned its framing from them. Publish your own honest comparison content and give the model a second source.
Where this fits in your broader brand audit
AI visibility isn't a separate discipline. It's downstream of everything else: positioning clarity, message consistency, review strength. Brands that are confused on their homepage and contradicted in their reviews are confused and contradicted in AI answers too, just with less chance to argue back.
That's why BrandAudit treats search and discovery presence, including how brands surface in AI assistants, as one of six intelligence layers in every report. Drop in your URL and it reads your website messaging, social content, customer reviews, and competitor signals alongside your discovery presence, then scores it all against eight strategy frameworks with up to five competitors benchmarked. You see whether your weak AI visibility traces back to a positioning problem, a review problem, or a consistency problem, and the 90-day roadmap tells you what to fix first. The free sample reports show what that looks like for real brands.
The answer box is the new shelf
For decades, brands fought for shelf position, then for search position. Now they're fighting for a sentence in an answer, and most don't know the fight is happening. The brands that win it early will look, in hindsight, like the ones that took SEO seriously in 2005.
Run the manual prompts this week. An ai visibility audit costs you an afternoon, and what you find, present or absent, accurate or stale, will tell you more about your brand's discoverability than your traffic report does.
Then get the full picture. Browse the free sample reports to see how discovery presence gets scored alongside positioning, reviews, and competitors, and run your own audit before your next buyer asks an assistant who to choose.
To see what these checks look like in a finished report, open the retail brand audit sample - every section is real and free to read.
Tags
