An AI visibility audit measures how often AI engines name your brand in answers to buyer questions. Run 60 to 80 category, problem and comparison prompts across ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews, divide the prompts that name your brand by the prompts you ran, then review which sources were cited and whether each mention is accurate. The audit takes two to four hours and gives you a baseline Citation Share.
Updated October 17, 2026. Part of the AI Communications pillar. Cluster: Citation Share Is the New Discoverability KPI · The 35-Prompt Citation Share Audit · Nonprofit AI Answers · The 5W AI Visibility Index · AI Communications Dictionary
You cannot improve what you have not measured, and that applies to AI visibility. Before a brand can improve its Citation Share, it needs to know what that share is today.
Most brands have not measured it. Most do not know where they stand inside ChatGPT, Claude, Perplexity, Gemini, or Google AI Overviews. The five steps below produce a baseline in two to four hours.
What does an AI visibility audit measure?
An AI visibility audit has two halves. The countable half asks how often the brand surfaces, and in what position, across a defined set of buyer-intent prompts. The qualitative half asks whether the description is accurate, whether the framing is one the team would have chosen, and whether a competitor is named first. A number alone does not show whether an AI tool is recommending a brand or undercutting it.
| Step | What you do | What you get |
|---|
| 1. Prompt inventory | Write 60 to 80 category, problem and comparison prompts | A reusable query library |
| 2. Run the prompts | Run each prompt on all five engines and log the answers | A results spreadsheet |
| 3. Baseline Citation Share | Divide prompts naming your brand by prompts run | A percentage by engine, query type and overall |
| 4. Source attribution | Record which sources each answer cites | Your retrieval anchors and your competitors' source map |
| 5. Characterization accuracy | Check category, capabilities and sentiment in each mention | A list of errors to fix at the source |
Step 1: How do you build a prompt inventory?
A prompt inventory is a structured set of the questions buyers in your category actually ask AI engines, written as category and problem queries instead of branded ones. Use the 35-prompt starter set if you want a pre-built inventory, since the buckets below cover the same query types.
Start with 60 to 80 prompts across three buckets.
Category queries. "What are the leading [category] firms?" "Who are the best [discipline] agencies?" "What companies do [specific function]?" These show which brands the engine sees as category leaders.
Problem and solution queries. "How do I [solve specific problem]?" "What is the best approach for [challenge]?" "Which brands are known for [capability]?" These show which brands are tied to expertise in buyer pain points.
Comparative queries. "[Brand A] vs [Brand B]." "What is the difference between [approach X] and [approach Y]?" These show how engines describe your brand next to competitors.
Run the full inventory, or a representative 25-prompt subset for a faster audit, on ChatGPT, Claude, Perplexity, Google AI Overviews and Gemini. Record platform, prompt, brands mentioned, whether your brand was mentioned (yes or no), how your brand was described, and the sources cited.
The answers differ by platform. A brand that dominates Perplexity may be absent from ChatGPT, and Google AI Overviews draws on a different source mix than Claude. The combined view across all five is the one to use for strategy.
Step 3: How do you calculate baseline Citation Share?
Citation Share equals the number of prompts where your brand is mentioned, divided by the total prompts run, multiplied by 100. Calculate it by platform, by query category and overall, because the per-platform breakdown usually shows where the biggest gaps are.
For the full definition and the weighted scoring formula, see Citation Share Is the New Discoverability KPI. For context on what competitive Citation Share looks like, the AI Platform Citation Source Index 2026 maps which of 50 domains drive citation, and the Who Controls AI Answers franchise tracks Citation Share by category.
Step 4: How do you audit source attribution?
For every prompt where your brand appears, note which sources the engine cites. This is the most useful layer of the audit, because it shows which earned media placements, research publications or owned pages actually drive AI retrieval.
Sources that appear across many prompts are your current retrieval anchors. Sources that should drive retrieval but do not point to gaps in coverage quality or source authority.
For every prompt where your brand does not appear, note which brands do and which sources support them. That competitor source map is the starting brief for your next earned media program. If your category has a published Citation Share Index, its Tier 1 publications are the highest-leverage targets.
Step 5: How do you check characterization accuracy?
Check each mention on three dimensions: category positioning, capability representation and sentiment.
Category positioning. Does the engine place your brand in the right category with the right positioning, or does it describe you as something adjacent, outdated or imprecise?
Capability representation. Does the engine describe your core capabilities correctly? A hallucination, where the engine states something false with confidence, is a citation risk that needs entity work to fix.
Sentiment. Does the engine describe you as a leader, a niche player, a historical reference or a primary recommendation? That framing shapes how buyers respond when they see your brand in an answer.
What are the three measurement layers?
An AI visibility audit is one of three distinct measurements. Teams that confuse them measure the wrong thing and act on the wrong brief.
| Measurement | Question it answers | Where it lives |
|---|
| AI Visibility Audit (brand-side) | How does my brand appear in AI answers, and how accurately? | This page |
| AI Visibility Index (entity-side) | Which named entities in a category do the engines cite, and in what rank order? | The 5W AI Visibility Index |
| Citation Share Index (publication-side) | Which publications do the engines pull from to build the answer in the first place? | The Citation Share Index |
Run the audit to find the gap. Read the Index for your category to find where the placement work closes it.
What should you do with the audit results?
A completed audit produces three outputs: a baseline Citation Share by platform and query category, a source attribution map, and a characterization accuracy assessment. The baseline is the number all later measurement compares against. The source map shows which placements drive citation and where the competitive gap sits. The accuracy assessment shows where the engine's model of your brand is wrong and where entity development is needed.
Read the result as a map, not a single grade. It shows where the brand is strong, where a competitor owns the answer, and where an AI tool is wrong about the brand. Those are three different problems with three different fixes: a strong position to defend, a competitive gap to close, and an inaccuracy to correct at the source.
These three outputs define the first 90 days of an AI Communications program. Run the audit quarterly, because Citation Share moves as competitors build and lose coverage, as new placements enter the retrieval pool, and as engines update their understanding of your category.
How do you track AI visibility over time?
An AI visibility tracker is any repeatable process that runs the same prompt set on a schedule and records whether the brand is named and which sources are cited. Two approaches exist: a spreadsheet run by hand, and a monitoring tool that runs prompts automatically.
| Approach | Best for | Limit |
|---|
| Manual spreadsheet | A first baseline and small prompt sets of 25 to 50 prompts | Slow to repeat, so runs tend to happen quarterly |
| Monitoring tool | Larger prompt sets and monthly or weekly runs across several engines | Each tool defines its own prompt pool and scoring, so numbers from different tools are not comparable |
Vendors in this space also publish research. Ahrefs used its Brand Radar product to find brand mentions in AI responses for its study of 75,000 brands, and Muck Rack's Generative Pulse product produces the data behind its What Is AI Reading? reports. Read vendor studies as useful evidence from a party that sells the tracking, and compare their method notes before relying on a number.
Whichever approach you use, record the date of every run and repeat each prompt more than once. Muck Rack's report states that the behaviors it observed may shift as models are updated or retrained, and EPR's AI citation volatility analysis reports a Digital Authority Partners finding that only 10.6% of cited URLs persisted across a 28-day window. A single run is a sample, and a month-to-month change can reflect an engine update instead of your work. The share calculation and the reporting method are laid out in the PR and SEO plan.
How do you check brand mentions in Gemini and ChatGPT?
Check brand mentions in Gemini and ChatGPT by running a fixed set of buyer prompts in a fresh chat for each prompt, then recording whether the brand appears and which sources the answer cites. The steps below take about an hour for 10 prompts on two engines.
- Write 10 category and comparison prompts that real buyers would ask, with no brand name in them.
- Open a new chat for each prompt, so earlier answers do not shape the next one.
- Record whether your brand is named, how it is described, and which sources the answer links.
- Run the same prompts on three different days and note which results change.
- Repeat for the two closest competitors, so the result has a comparison point.
Gemini and ChatGPT differ in what they cite, so keep their results in separate columns. The ChatGPT Citation Source Index and the AI citation statistics for 2026 show which domains each engine cites most, which tells you where a mention would need to appear for the answer to name you.
How does your audit number compare with your category?
The audit gives you your own number, and published research on the same five engines tells you whether that number is good.
The 5W AI Visibility Index scores every named entity in a category against the same grid, and the full methodology is here. Live indexes: Defense & Aerospace · Sports Betting & Gaming · Legal · Legal Tech · Credit Cards · Cannabis · Health & Wellness · Lottery · Amusement Parks · Pet Industry · Interior Design & Architecture · Restricted Categories · Bollywood · South Florida Luxury Real Estate.
Applied audits of full categories: Medical Aesthetics · Nonprofits.
Publication-side companions, which show who the engines pull from: the AI Platform Citation Source Index 2026, Who Controls AI Answers, and the trade-press citation indexes for Alcohol & Spirits and Crisis Communications.
Every study in one place: Everything-PR Research: The Master Index. Eight research series, updated continuously.
Related: Citation Share Is the New Discoverability KPI · The 35-Prompt Citation Share Audit · Measurement & Reporting: The 12-Month Playbook · What Is AI Communications? · AI Visibility Hub · AI Platform Citation Source Index 2026 · Everything-PR Research: The Master Index · AI Communications Dictionary
