AI Visibility Audit
See where AI answers mention and cite your brand today, by engine, model and date.
Our AI visibility audit shows where ChatGPT, Perplexity, Gemini and Google AI Overviews mention or cite your brand today, which sources they lean on, and what to fix first. Every result comes from repeated runs, recorded by engine, model and date. It's an AI search audit service with a written report, not a single score.
What you get:
- Prompt-level results by engine, model and date
- Entity, technical and source checks, with evidence
- A prioritized list of what to fix
What the audit covers
An AI visibility audit, or generative search audit, checks four things: how engines answer a set of prompts about your market, whether your company data is consistent, whether crawlers can read your site, and which sources the answers cite.
#baseline-measurement
Baseline measurement
We run an agreed set of prompts through ChatGPT, Perplexity, Gemini and Google AI Overviews / AI Mode. We settle the engine set when you request the audit. For each engine we record the model version, where the platform shows it.
We build the prompt set with you before the first run: category, comparison, problem-to-solution and branded prompts. Each prompt runs more than once, because answers change, and every run carries its date. You learn how often your brand is named, and how much the answers vary.
We agree the scope and timing of the audit with you before the first run.
#brand-entity
Brand entity
We check whether the basics about your company match across your site, profiles and directories: name, what you do, who it's for, where you operate. We also record what the engines say about you.
You learn where the facts disagree. Inconsistent entity data goes into the report as an item to correct at the source: on your site and in your profiles.
#technical-access
Technical access
We check whether the search crawlers of the engines and Googlebot can read your site. That covers robots.txt rules for OAI-SearchBot, PerplexityBot and Googlebot, indexing, canonical tags, noindex, your sitemap, and whether the main content is in the HTML without depending on JavaScript. Whether llms.txt affects AI visibility is still an open question; our llms.txt AI visibility test design shows how to check it.
What the official documentation says:
- Google has no extra requirements for AI features: a page needs to be indexed and eligible to show with a snippet. See Google's guidance on AI features.
- Sites that opt out of OAI-SearchBot aren't shown in ChatGPT search answers, though they can still appear as navigational links. See OpenAI's crawler documentation.
- PerplexityBot is used to surface and link websites in Perplexity search results. See Perplexity's crawler documentation.
GPTBot and Google-Extended are separate settings for AI model training (for Google-Extended, Gemini training and grounding). Google says Google-Extended doesn't affect inclusion in Google Search. Allowing or blocking them is your choice. The report records your current setting and doesn't tell you to close it. See OpenAI's crawler documentation and Google's crawler documentation.
#cited-sources
Which sources the answers cite
We list the sources the engines cite for your prompt set: your own pages, competitors' pages, industry publications, directories and catalogs, review sites, documentation and public discussions.
You learn which of those sources relate to you and where honest work is possible. We don't claim to know why an engine picked a source: the platforms' documentation doesn't describe that. Several of those links can lead back to one original page: see AI citation source lineage.
How the audit works
Five steps, from scope to walkthrough.
- 1
Scope
Your brand, main competitors, markets, languages and engines. We agree the prompt set with you.
- 2
Run
Each prompt runs several times in each engine in scope. We record the engine, model and date.
- 3
Check
We review brand data, technical access and cited sources, with evidence: URLs, robots.txt lines, check dates.
- 4
Report
You get the report in the structure below, plus a list of priorities.
- 5
Walkthrough
We go through the report with you: what we found and what to do first.
What you get: the audit report
The report has six parts. For each one, the tables below show the columns and what goes in them. This is the structure, not sample data: every value in your report comes from your own runs.
Report sample · prompt-level resultsstructure only · no real data
Rows fill in only from real runs.
| Prompt | Type | Engine | Model | Run date | Run | Brand mentioned | Your page cited | Competitors |
|---|---|---|---|---|---|---|---|---|
| [prompt P-01] | [type] | [engine] | [model] | [date] | 1/N | no data | n/a | n/a |
| [prompt P-01] | [type] | [engine] | [model] | [date] | 2/N | no data | n/a | n/a |
| [prompt P-02] | [type] | [engine] | [model] | [date] | 1/N | no data | n/a | n/a |
#report-prompt-results
Prompt-level results
One row per run.
Column
What it shows
Prompt
The exact prompt text, as agreed before the first run
Prompt type
Category, comparison, problem-to-solution or branded
Engine
ChatGPT, Perplexity, Gemini, Google AI Overviews / AI Mode, whichever were in scope
Model
The model version, where the platform shows it; otherwise "not shown"
Run date
The date of this run
Run
Which repeat of the prompt this is
Brand mentioned
Yes or no: your brand is named in the answer
Your page cited
The URL of your page, if the answer links to it
Other sources cited
URLs the answer links to that aren't yours
Competitors mentioned
Which of the agreed competitors are named
What the answer says about you
A short note; inaccurate facts are flagged
#report-summary
Prompt summary
One row per prompt and engine.
Column
What it shows
Prompt
The prompt this row summarizes
Engine
The engine the runs were made in
Runs
How many times the prompt ran in this engine
Runs with a mention
In how many of those runs your brand was named
Runs with your page cited
In how many of those runs the answer linked to your page
Competitors named most often
The agreed competitors that appeared most across the runs
Notes on variation between runs
How much the answers differed from run to run
These are counts across runs within the audit's date window. They're not a forecast and not a position.
#report-entity
Entity check
One row per fact.
Column
What it shows
Fact
A basic fact about your company: name, what you do, for whom, location and similar
On your site
What your site says
On profiles and directories
What your profiles and directory listings say
In AI answers
What the engines said about it in the runs
Consistent?
Whether these sources agree
Where to correct it
The source where the fix belongs
#report-technical
Technical access checks
One row per check.
Column
What it shows
Check
What we tested: robots.txt rules per bot, indexing, canonical, sitemap, content without JavaScript
What we found
The result of the check
Evidence
The URL, robots.txt line, HTTP status and check date behind the result
Priority
How soon it's worth fixing
#report-sources
Cited sources
One row per source.
Column
What it shows
Source
The domain the answers cite
Source type
Your site, competitor, publication, directory, review site, documentation or discussion
Cited for (prompts and engines)
Which prompts and engines cited it
Mentions you?
Whether the source mentions your brand
Honest options
What you can do about it: correct your profile, publish primary information, take part with disclosure, or do PR with a real reason
#report-priorities
Priorities
One row per action.
Column
What it shows
Action
What to do
Why (link to the finding)
The finding in the report that the action responds to
Who does it (us or you)
Whether your team or ours would carry it out
Priority
What to do first
Who the audit is for
The audit fits if any of these sounds like you:
- You don't know whether AI engines name your brand.
- You see engines naming your competitors, and not you.
- You want a starting point before any work begins, whether your team's or ours.
What the audit can show, and what it can't
The audit is a snapshot of specific dates. AI answers change between runs, models and dates, so we report what we saw, not what will stay true.
We do
- Show where you're mentioned and cited for the agreed prompts, engines and dates.
- Show which sources the answers lean on.
- List entity and technical issues, with evidence.
- Give a prioritized list of what to fix first.
We don't promise
- That fixing the list will get you named, cited or recommended by any engine.
- A spot in any AI answer, or a position in Google.
- That results stay the same after the audit dates.
Acting on the list may improve your chances, but we can't say whether or by how much. After that, we measure and report.
We don't use
Hidden paid mentions, fake reviews or "independent" articles, fake accounts, spam in discussions, hidden text or prompt injection aimed at models, or mass-produced pages written for prompts. None of our recommendations rely on them.
Tally · same prompt, 8 runs
[engine] · [model] · [date range]Answers change from run to run, so we count repeats instead of trusting one screenshot.
Want to do it yourself? Read how to measure AI visibility yourself.
How we measure in the audit
One answer proves nothing. This is the protocol the audit follows.
- 1
A fixed prompt set
We agree the prompts with you and don't change them after the first run.
- 2
Repeats, not one answer
Every prompt runs more than once, because answers change from run to run.
- 3
Engine, model, date on every row
Each run in the report shows its engine, its model (if the platform shows it) and its date.
- 4
Mentions and citations counted separately
Whether your brand is named and whether the answer links to your page sit in different columns (see ChatGPT brand citation mismatch).
- 5
Counts, not screenshots
The result is a count across all runs in the audit's date window, not one screenshot.
- 6
Repeatable later
The same prompt set can be run again and compared. That's what ongoing AI visibility tracking does.
What we need from you
- Your brand and website
- Your main competitors
- Your market and language
- The products or topics that matter most
Request baseline audit
Start here. Tell us your brand, competitors and market. We confirm the scope and the prompt set with you before the first run, and we don't run anything until you've agreed.
The price is on request. We agree the scope and timing with you before the first run.
After the audit
The report is yours. You can work through the priorities with your own team. If you'd like us to do the work, see what we do and generative engine optimization. To re-run the same prompt set on a schedule, see ongoing AI visibility tracking.
Frequently asked questions
What is an AI visibility audit?
An AI visibility audit is a one-time check of where AI engines mention and cite your brand. It covers four parts: a baseline measurement across repeated prompt runs, your brand data, technical access for crawlers, and the sources the answers cite.
Can you guarantee our brand will show up in AI answers after the audit?
No. The audit shows the current picture and what to fix first. Answers change between runs, models and dates, so no one can promise a result. We work to improve your chances, then measure and report. See what the audit can show and what it can't.
Which AI engines does the audit cover?
ChatGPT, Perplexity, Gemini, and Google AI Overviews and AI Mode. We agree the set with you when you request the audit.
Can you run a ChatGPT visibility audit only?
Yes, we can agree that scope. Keep in mind that results from one engine don't carry over to the others. For work on that engine, see ChatGPT SEO.
Why do results differ between runs?
AI answers aren't fixed. The same prompt can return a different answer on another run, in another model or on another date. That's why we run each prompt more than once and log the date. See how we measure.
What do you need from us?
Your brand and website, your main competitors, your market and language, and the products or topics that matter most.
If the audit needs any access on your side, we agree it with you first.
How long does the audit take?
It depends on the number of engines, the size of the prompt set, and the markets and languages. We confirm the scope before the first run.
How much does an AI visibility audit cost?
Price is on request. It depends on the number of engines, the size of the prompt set, markets and languages, how often you re-measure, and how much content and technical checking is in scope.
What happens after the audit?
You get the report and a walkthrough. You or we can then do the work: see after the audit.
Start with a baseline
See where you are today, then decide what to do about it.