One audit. Every engine. A number you can act on.
You can't fix what you can't see, and right now you can't see what AI tells your buyers. The AI Visibility Audit makes it visible, scores it, and hands you the fix list. Repeatable, so when we fix it, you can watch the number move.
What we measure (and why each one protects revenue)
Every major engine
ChatGPT, Claude, Gemini, and Perplexity. Queried the way real buyers ask: branded, unbranded, comparative, geographic. So you know: where you're named, where you're invisible.
Web-on and web-off
Each engine runs twice: with live web access, then without. So you know: how much of your visibility is durable, and how much vanishes with one lost citation.
A training-data probe
A locally-run model, web disabled, reveals what AI actually memorized about you. So you know: the stale facts and made-up claims AI states as truth when it doesn't look you up.
Competitor share-of-voice
Who the engines name in your category answers, and how often. So you know: exactly which competitors own the shortlist, and where they're beatable.
Name and brand-safety check
Does your bare name resolve to you, or drift to a bigger namesake? Are there false claims circulating? So you know: before a customer does.
Your AI Visibility Index
A weighted 0-100 score per topic: recall, accuracy, share-of-voice, citations. So you know: your baseline, your priorities, and later, your proof of lift.
Every finding lands in one of five states, and each state tells you the move
No 80-page report you'll never read. Every brand fact gets one of five labels, and each label maps to one action:
| State | What's happening | What we do about it |
|---|---|---|
| Baked-in | AI knows it and repeats it live | Defend it. Don't assume it's permanent. |
| Retrieval-dependent | Only visible while a live citation holds | Harden the sources so it survives. |
| Confusion-corrected | AI gets it wrong until the web bails it out | Reinforce entity signals so it's never wrong. |
| Persistent gap | Invisible everywhere | Build net-new presence. Highest leverage. |
| Regression | AI knew it and lost it | Find what changed. Recover it. |
Then we fix it. In this order, because this order works.
Make your brand unmistakable to machines
Schema markup, an accurate knowledge-graph entity, consistent details everywhere. This is what stops a same-named company from stealing your answers and overwrites AI's stale memory at the root.
Correct the record where AI actually reads
Encyclopedias, directories, review sites, forums. The third-party sources engines trust. Fixing your own website alone doesn't cut it.
Publish pages machines can quote
An llms.txt, on-page Q&A, structured data, aimed at the categories your buyers actually ask AI about. So engines extract you accurately instead of guessing.
Earn the sources AI cites
Engines pull from a predictable set of authority domains. Getting placed there is what flips "not mentioned" into "recommended."
Re-audit and show you the lift
Same queries, same scoring, same baseline. You see recall, accuracy, and share-of-voice move in black and white. No faith required.
Why "re-auditable" is the part that protects you
Anyone can tell you AI matters. Almost nobody can hand you a number today, do the work, and show you the same number moving in 90 days. That's the difference between buying a service and buying a story. If you've been burned by marketing spend you couldn't verify, this is the fix: a baseline that's measured, not vibes.
Your move
Get your baseline before your competitor gets theirs
Every week you wait, AI keeps answering your buyers without you, and whoever moves first gets cited first. One audit gets you the number, the gaps, and the plan.