AI Visibility Audit - Does ChatGPT Recommend Your Brand?
You know your Google ranking. But do you know whether ChatGPT, Perplexity and Google AI Overviews recommend your brand - or your competitors? More and more buying decisions start with a question asked to AI, not a phrase typed into a search box. An AI visibility audit (a GEO audit) answers the question: do you even exist in generative answers, who cites you, and how do you compare to competitors? This guide explains how to run one - including the new reports that arrived in 2026.
You know your Google ranking - but do ChatGPT, Perplexity and AI Overviews recommend you or your competitors? An AI visibility audit (GEO) measures AI Share of Voice, citations and sentiment. I explain how to run one step by step.
What changed in 2026: the first official data
Until recently there was no platform data on AI visibility at all. That just changed - but only partly:
- Google Search Console - the "Generative AI" report (June 2026): a dedicated breakdown of impressions in AI Overviews and AI Mode, by page, country and date. Important limitation: it's impressions only - no clicks, CTR or query data. Google stresses this data was always included in the overall Performance report totals; the new part is the isolated view. An opt-out toggle to exclude content from AI features also arrived.
- Bing Webmaster Tools - the "AI Performance" report (preview since February 2026, expanded in June): shows when your site is cited in Microsoft Copilot answers and Bing AI summaries - citation counts per URL, "grounding queries" (the phrases the AI used to retrieve content), and since June also intents, topics, citation share and comparisons.
Still, the picture is incomplete: ChatGPT and Perplexity provide no first-party analytics, and Google's data is impressions only. So a GEO audit still requires actively querying the engines.
/// THE FIRST OFFICIAL AI DATA (2026)
- ›Impressions in AI Overviews and AI Mode
- ›Breakdown: page / country / date
- ›NO clicks, CTR or queries
- ›Citations in Copilot / Bing AI
- ›Grounding queries, intents, topics
- ›Citation share + comparisons
* ChatGPT and Perplexity still have no first-party analytics — a GEO audit still requires active querying.
What an AI visibility audit measures
The heart of the audit is AI Share of Voice - the share of AI answers (across a defined set of your customers' questions) in which the brand is mentioned, cited or recommended, relative to competitors.
| Dimension | The question it answers |
|---|---|
| Presence | Does AI mention your brand at all? |
| Share of Voice | How often, compared with competitors? |
| Citations | Which of your pages are given as a source? |
| Sentiment | In what context is your brand mentioned - positive or not? |
| Factual accuracy | Does AI tell the truth about you, or "hallucinate"? |
| Engine coverage | ChatGPT vs Perplexity vs AI Overviews vs Copilot vs Gemini |
How to run an audit - step by step
- 1.Build a question set. 30-100 real customer questions (problem, comparison, "best X for Y", local, branded).
- 2.Query the key engines with the same questions: ChatGPT (with search), Perplexity, Google AI Overviews and AI Mode, Copilot, Gemini.
- 3.Sample repeatedly. Model answers are non-deterministic - the same question yields different answers. Ask several times to average out.
- 4.Record results: was the brand mentioned, was it cited (and which page), sentiment, are the facts correct.
- 5.Compute Share of Voice against 2-3 competitors.
- 6.Diagnose gaps - where competitors are cited and you aren't, and why.
How to build a good question set
An audit is only as good as the questions you ask. A set of 30-100 questions should mirror the customer's real decision journey, not your SEO keywords. Six categories worth covering:
| Question category | Example | What it checks |
|---|---|---|
| Problem | "how to automate invoicing in a small business" | Whether AI associates you with the problem you solve |
| Comparison | "X or Y - which is better for…" | Whether you appear in head-to-heads with competitors |
| "Best X for Y" | "best tool for…" | Whether you're on shortlists of recommendations |
| Alternatives | "alternatives to [competitor]" | Whether AI suggests you as a replacement |
| Local | "AI consultant in [city]" | Visibility in queries with local intent |
| Branded | "what does [your brand] do" | Whether AI tells the truth about you (factual accuracy) |
The rule: ask in the customer's language, in full sentences, the way people actually talk to AI - not keywords. The closer to real buying intent, the more useful the audit.
Where engines get sources - and why Bing is key
This is the audit's most important technical takeaway. Engines pull from different indexes:
- Google AI Overviews and AI Mode → Google's index (RAG over the normal index, with "query fan-out").
- ChatGPT and Microsoft Copilot → largely the Bing index. Analyses of ChatGPT citations show ~87% overlap with Bing's results.
- Perplexity → a hybrid model: its own crawler (PerplexityBot) plus search-engine APIs.
The takeaway: you can't do GEO while ignoring Bing. Being indexed and clean in Bing Webmaster Tools unlocks ChatGPT and Copilot simultaneously - yet most companies focus only on Google.
/// WHERE AI ENGINES GET THEIR SOURCES
This is why Bing indexation unlocks ChatGPT and Copilot at once
How to read the results - from raw data to a plan
Raw "mentioned / not mentioned" is just the start. The audit's value lies in prioritizing gaps:
- High-value gaps - questions close to the buying decision (comparisons, "best X") where competitors are cited and you aren't. Direct your effort here first.
- Easy-to-close gaps - questions you have content for but aren't cited on (often a technical problem: no Bing indexation, blocked bots, poor extractability).
- Factual errors - AI says something untrue about you. These are urgent, because they damage your image on every query.
A practical matrix: X axis = question value (how close to a decision), Y axis = difficulty of closing the gap. Start in the "high value / low difficulty" quadrant - quick wins, which are most often technical in nature.
Reverse-engineering competitors - why AI cites them, not you
When the audit shows a competitor cited for a given question, ask the simple question: why? The most common reasons, from easiest to fix:
- They're in the Bing index and you aren't - the most common and cheapest gap to close.
- They have extractable content - a direct answer, data, a table the model can "lift"; you have filler.
- They're mentioned externally - Reddit, media, rankings, Wikipedia; their entity is more strongly corroborated.
- They have fresher content - the model prefers current sources for date-sensitive questions.
Open the competitor's cited page and read it the way a model would: does it answer directly? Does it have numbers and quotes? Is it current? That's your to-do list.
Brand hallucinations - how to detect and correct them
The most dangerous audit result isn't "not cited" - it's AI saying something untrue about you: a made-up offer, a wrong price, a non-existent review, confusion with another company. The fix mechanism:
- Unify your own source of truth - a consistent, unambiguous "About" page and structured data; models learn from what you publish.
- Strengthen external corroboration - the more credible, consistent sources, the lower the chance of a hallucination.
- Monitor after the fix - re-run the same questions after a few weeks; models update their "understanding" of a brand with a lag.
Tools for tracking at scale
A whole category of AI Share of Voice tools has emerged: Profound, Otterly, Peec AI, AthenaHQ, Scrunch (acquired by Sitecore in 2026), plus modules in classic SEO tools - Semrush AI Visibility Toolkit and Ahrefs Brand Radar. They differ in methodology: some generate synthetic prompts (Semrush), some rely on real search-derived queries (Ahrefs). Prices, selection criteria and the "build your own" variant are compared in a separate post. On top of that come first-party reports: Bing AI Performance and the Generative AI report in GSC.
From audit to improvement
An audit is the diagnosis - the treatment is GEO. The most common findings and actions: open the site to AI bots (GPTBot/OAI-SearchBot, PerplexityBot, Bingbot) and ensure indexing in Bing too, rewrite key content for citability, build a consistent entity and earn third-party corroboration. Broader context: SEO vs GEO, getting cited in ChatGPT and Perplexity/SearchGPT.
A minimal one-day audit - where to start
You don't need to buy a several-hundred-dollar-a-month tool right away. You can run a first, manual audit in a single day:
- 1.List 20 questions your customers actually ask (problem, comparison, "best X for Y", branded).
- 2.Ask them in ChatGPT (with search), Perplexity and Google AI Mode - each question 2-3 times, because answers are non-deterministic.
- 3.Tally in a spreadsheet: was your brand mentioned, was it cited (and which page), which competitors appeared, what was the sentiment.
- 4.Compute a simple Share of Voice: in how many of the 20 questions you appeared vs competitors.
This quick pass shows whether you exist in AI answers at all and where competitors are beating you - enough to set priorities before you reach for scale monitoring tools.
The audit shows whether the brand appears in answers. If it turns out the models say something untrue about it, go to the post on brand reputation in AI - it covers where such errors come from and how to correct them.
---
I run AI visibility audits (GEO) as part of full AI-GEO - with a concrete improvement plan. I teach it in the SEO & GEO course. Get in touch - I'll check whether AI recommends you or your competitors.
Worth reading next:
/// RELATED_SERVICES
Need these concepts implemented? Explore the services related to this topic.
/// RELATED_RECORDS
Brand Reputation in AI - What to Do When ChatGPT Says Something Untrue About Your Company
In a July 2026 Searchable study, 93% of 1,704 brands had at least one fact wrong or missing in AI answers, and small businesses got fabrications far more often. Where models get untrue facts about a brand, how to audit them, five correction paths and what the law says when an error costs a contract.
Image SEO in the AI Era - Google Lens, Visual Search and Graphics Models Cite
Google Lens now processes close to 20 billion visual searches a month, and Circle to Search runs on over 580 million Android devices - images stopped being an afterthought to text and became a search surface of their own, with its own rules. Here's what alt text actually does (and doesn't do), how to wire images into ImageObject structured data, why an infographic built as HTML beats an infographic-as-image for citability, which file format to pick in 2026, and what the EU AI Act says about labeling AI-generated graphics.
Bing, Microsoft Copilot and IndexNow - the Forgotten Ecosystem Feeding AI Answers
On July 1, 2026, Fabrice Canel, Bing's longtime head of crawling and indexing and the chief architect of the IndexNow protocol, retired from Microsoft. It's a good moment to notice how much of today's AI-generated answers actually run on the infrastructure he built. The whole SEO world is debating Google AI Overviews, while Microsoft Copilot entirely, and ChatGPT substantially, pull sources from the Bing index - and since February 2026, Bing itself shows you when you're being cited. Here's how the ecosystem works, how to configure it, and how much traffic it actually delivers.
Signal received?
Terminate
Silence
Initiate protocol. Establish connection. Let's build something loud.