How To Run An AI Visibility Audit In 2026: A Step-By-Step Guide

Short answer: An AI visibility audit measures how often, how accurately and in what context AI assistants mention and cite your brand. You build a set of prompts your buyers actually ask, run them across ChatGPT, Google AI Overviews and AI Mode, Perplexity, Gemini and Copilot, and record whether you appear, where, how you are described and which sources the answer relies on. Then you check the technical basics that decide whether AI systems can read your site at all, and turn the gaps into a short list of fixes. A first audit takes one to three days for most brands.

Traditional SEO audits ask whether your pages rank. An AI visibility audit asks a different question: when someone asks an AI tool for a recommendation in your category, are you part of the answer? This guide gives you a repeatable process, the metrics to track and a scorecard you can reuse every quarter.

What Is An AI Visibility Audit?

An AI visibility audit is a structured review of your brand’s presence inside AI-generated answers. It covers four things:

• Presence: whether your brand is mentioned or linked for the prompts that matter to you.
• Accuracy: whether the AI describes your products, pricing, locations and positioning correctly.
• Sources: which pages, on your site and elsewhere, the AI relies on when it talks about you or your category.
• Access: whether the crawlers behind each AI product can actually reach and read your content.

It overlaps with an SEO audit but is not the same. Our explainers on GEO vs SEO and AEO vs SEO go deeper into how the disciplines differ.

Why Run One Now?

AI answers now sit in front of a large share of research journeys, especially for B2B services and considered purchases. When a buyer asks for “the best payroll software for a 50-person company” or “a reliable SEO agency in Denver,” the shortlist they see is decided by the AI, and brands that are missing never get the click.

There is also a measurement gap. Google’s Search Console added dedicated generative AI performance reports in June 2026, showing impressions for AI Overviews and AI Mode by page, country, device and date. That helps with Google, but it does not tell you what ChatGPT or Perplexity say about you. An audit fills that gap.

What You Need Before You Start

• A spreadsheet for prompts and results (the scorecard below gives you the columns).
• Access to the AI tools your buyers use. At minimum: ChatGPT with search, Google Search with AI Overviews and AI Mode, Perplexity and Gemini. Add Microsoft Copilot and Claude if they matter in your market.
• A list of three to five competitors you lose deals to.
• Access to Google Search Console, your server or CDN logs, and your robots.txt file.
• Optional: an AI visibility tracking tool to automate prompt runs. Our roundup of AI visibility tracking tools compares the main options.

Step 1: Build Your Prompt Set

The quality of the audit depends on the prompts. Aim for 50 to 150 prompts that mirror how real buyers ask questions, grouped by stage:

Prompt type	What it tests	Example
Category discovery	Whether you make shortlists	“Best AI SEO agencies for SaaS companies”
Problem and solution	Whether you appear for the pain you solve	“How do I get my brand mentioned in ChatGPT answers?”
Comparison	How you are positioned against rivals	“Profound vs Peec AI for brand monitoring”
Brand	Accuracy of what is said about you	“What does [Your Brand] do and how much does it cost?”
Local or vertical	Visibility in a niche or city	“Top law firm marketing agencies in Chicago”

Pull prompts from sales call notes, support tickets, “People also ask” boxes, Search Console queries and the questions customers ask on Reddit or in review sites. Write them in natural language, the way someone would type or speak to an assistant, not as short keywords.

Step 2: Run The Prompts Across Platforms

Run every prompt on each platform you chose and log the results. A few rules keep the data clean:

• Use clean sessions. Log out or use a fresh profile so personalization and memory do not skew results.
• Run each prompt more than once. AI answers vary from run to run. Three runs per prompt per platform is a sensible minimum, and tracking tools that sample many times give steadier numbers.
• Record the date, platform and mode. ChatGPT with and without search, or AI Overviews versus AI Mode, can return very different answers.
• Save screenshots or exports. You will want evidence when an answer is wrong.

Step 3: Score Each Answer

For each prompt and platform, record these fields:

• Mentioned: yes or no.
• Linked or cited: whether your site appears as a source.
• Position: first, top three, or later in the answer.
• Sentiment and framing: positive, neutral or negative, and how you are described.
• Accuracy: any wrong facts about pricing, services, locations, leadership or availability.
• Competitors mentioned: which rivals appear alongside or instead of you.
• Sources cited: every URL the answer links to.

From that log you can calculate the core metrics:

Metric	How to calculate it	Why it matters
Mention rate	Prompts where you are mentioned ÷ total prompts	Your basic presence in AI answers
Citation rate	Prompts where your site is a cited source ÷ total prompts	Whether your own content is trusted as evidence
Share of voice	Your mentions ÷ all brand mentions across the prompt set	Your presence relative to competitors
Top-three rate	Prompts where you appear in the first three brands ÷ prompts where you appear	Whether you are a lead option or an afterthought
Accuracy rate	Answers with no factual errors about you ÷ answers that mention you	Risk to reputation and conversions

Step 4: Check Technical Access

If AI crawlers cannot reach your pages, nothing else in the audit will improve. Check these items:

• robots.txt rules for search and retrieval bots. OpenAI’s crawler documentation says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers. Perplexity’s crawler documentation recommends allowing PerplexityBot. For Google, AI Overviews and AI Mode use normal Googlebot crawling, and a page must be indexed and eligible for a snippet.
• The difference between training and search bots. GPTBot collects data for model training, while OAI-SearchBot powers ChatGPT search. Blocking one does not block the other, so make sure your rules match your intent.
• CDN and firewall settings. Bot protection at the edge can quietly block AI crawlers even when robots.txt allows them. Check your logs for requests from the crawlers you care about and for 403 responses.
• Rendering. Key facts such as prices, service areas and product details should be in the HTML text, not hidden behind scripts, tabs or images.
• Snippet controls. A stray nosnippet or max-snippet directive can keep pages out of Google’s AI features.
• Indexing. Use Search Console to confirm your most important pages are indexed.

Files like llms.txt are optional. Google says you do not need them for its AI features, as covered in our llms.txt guide.

Step 5: Map The Sources AI Relies On

AI answers are built from retrieved pages, and for commercial queries many of those pages belong to someone else. Take every cited URL from Step 3 and group it:

• Your own site: which pages get cited, and which important pages never do.
• Competitor sites: where rivals are earning citations you are not.
• Third-party lists and reviews: “best of” roundups, comparison sites, review platforms and industry directories.
• Community and reference sites: Reddit, YouTube, Wikipedia, LinkedIn and forums.
• News and trade media: coverage that shapes how your category is described.

The result is a source gap list: the specific domains and pages where competitors appear and you do not. That list often becomes the most valuable output of the audit, because it shows exactly where digital PR, outreach and list placements will have the most effect. Our guide to what a brand mention is for SEO explains why unlinked mentions matter here too.

Step 6: Review Your Own Content For Extractability

For the prompts you care about most, look at the page on your site that should answer each one and ask:

• Does the page answer the question in the first two or three sentences?
• Is each section built around a clear question, with a heading that matches how buyers ask?
• Are prices, specs, locations and other facts stated plainly in text, ideally in a table?
• Is the page current, with a visible updated date and no outdated claims?
• Is your company described consistently across your site, social profiles, directories and review sites?

Google’s query fan-out, which it describes on its AI features page, splits one question into many related searches. Pages that cover the sub-questions behind a prompt, such as pricing, alternatives and how long something takes, have more chances to be pulled in. Our guide to ranking in AI Overviews covers this in detail.

Step 7: Turn Findings Into A Prioritized Action Plan

Sort every issue into one of four buckets and rank by impact and effort:

1. Access fixes: blocked crawlers, unindexed pages, snippet restrictions. Usually quick and high impact.
2. Accuracy fixes: wrong facts in AI answers. Correct the source pages, update directory and review profiles, and publish clear factual pages that AI tools can cite.
3. Content fixes: missing pages for high-value prompts, weak answers, no comparison or pricing content.
4. Authority fixes: absence from the third-party sources AI relies on. These take longest but often matter most.

Set a baseline from this audit and re-run the same prompt set monthly or quarterly so you can show change over time.

AI Visibility Audit Scorecard Template

Use this as the summary page of your audit:

Area	What to record	Baseline	Target
Mention rate	Across all prompts and platforms	e.g. 18%	30% in two quarters
Citation rate	Your domain cited as a source	e.g. 6%	15%
Share of voice	Versus your top competitors	e.g. 9%	15%
Accuracy	Answers with no errors about you	e.g. 70%	95%
Crawl access	Search bots allowed and not blocked at the CDN	Pass or fail	Pass
Source gaps	Top third-party domains citing competitors but not you	List of 10 to 20	Half covered
Content gaps	High-value prompts with no strong page on your site	List	All covered

Doing It Yourself Or Hiring Help

A small team can run a solid first audit manually with a spreadsheet and a few days of work. It gets harder to sustain at scale, because answers change frequently and you need many runs per prompt to see real trends. That is where tracking tools and specialist agencies come in. If you want outside help, our lists of the best GEO agencies and top AI SEO agencies compare firms that offer AI visibility audits and ongoing programs. For tools, see our comparison of AtomicAGI, Profound and Peec.

Frequently Asked Questions

How often should I run an AI visibility audit?

Run a full audit once or twice a year and a lighter tracking check monthly. Re-run the full audit after a major site change, rebrand, product launch or reputation issue.

How many prompts do I need?

Between 50 and 150 prompts is enough for most brands. Large companies with many product lines may need several hundred, split by product or region.

Which AI platforms should I include?

Start with ChatGPT, Google AI Overviews and AI Mode, Perplexity and Gemini. Add Copilot, Claude or regional assistants if your buyers use them.

Can I see AI visibility in Google Search Console?

Partly. Search Console’s generative AI performance report shows impressions from AI Overviews and AI Mode, but it does not cover ChatGPT, Perplexity or other assistants.

What if AI tools say something wrong about my company?

Fix the facts at the source: your own pages, your Google Business Profile, directories, review sites and any outdated articles. AI answers usually reflect what the web says. If the problem is reputational, our lists of AI Overviews reputation management companies and ChatGPT reputation management companies cover firms that specialize in it.

The Bottom Line

An AI visibility audit gives you a baseline for a channel most companies are still guessing about. Build a realistic prompt set, test it across the platforms your buyers use, check that AI crawlers can reach you, and map the sources that shape the answers. The fixes usually come down to access, accuracy, better answers on your own pages and more presence on the third-party sites AI trusts. To go deeper on two of the platforms that cite sources most visibly, read our guide to ranking in Perplexity and Google AI Mode.