What an AI Brand Audit actually finds

"Audit" is an overloaded word. Most things sold under it are either a vibes report — a consultant's impressions dressed up as findings — or an SEO scan pointed at a new acronym. The AI Brand Audit is neither. It is a week-long instrument that produces three specific, defensible outputs about how AI search treats your brand. Here is exactly what comes out the other end, so you know what you're buying before you book it.

One — your share of model, in numbers

The first output is a measurement, not an opinion. We ask the assistants the questions a real buyer would ask in your category — across the spread of phrasings people actually use — and record who gets named, how often, and how they're described. From that we compute your share of model: the percentage of relevant AI recommendations in which your brand appears, set against the competitors who appear in your place.

It is deliberately reproducible. You get the raw assistant responses, not just our summary, so you can re-run the prompts and check the number yourself. The point of leading with a hard figure is that it ends the argument about whether the recommendation moved — you can see your own answer.

Two — the architectural cause of the gap

A number alone isn't actionable; you need to know why. So the second output diagnoses the cause, and it is almost always architectural rather than editorial. Common findings:

• Inconsistent facts. Your products, dosages and claims say different things across your site, retailers and reviews, so the model hedges and drops you from the shortlist.
• No machine-readable structure. The facts exist but live in prose and PDFs, so an engine has to infer them — and inference is where you get described wrong or skipped.
• An unmodeled catalog. The relationships that make your answers correct — this product, this dose, this contraindication — aren't encoded anywhere a machine can traverse, so a text-similarity engine guesses and gets the specifics wrong.
• Thin corroboration. You only describe yourself, in one place, in your own words, which reads to the model as a single unverified claim.

Naming the real cause is what separates this from a checklist. The fix for "inconsistent facts" is nothing like the fix for "unmodeled catalog," and spending on the wrong one is how brands waste a year.

Three — a roadmap ranked by ROI, that you keep

The third output is the plan: the specific systems that would close your gap — a brand ontology, a visibility engine, a content engine, an agent workflow — ranked by expected return, not by what we'd most like to sell. If two cheap fixes get you most of the way, the roadmap says so and puts them first.

Crucially, the roadmap is yours to keep. It's written to be actionable by any competent team, including your own. Most clients choose to build the top item with us as a System Sprint and then have us operate it — but that is a decision you make after you hold the diagnosis, not a string attached to it.

Why it's the front door to everything

We make the audit a flat $1,500 and lead with it for one reason: it aligns us with you before a dollar of build is spent. You get a measured, honest read and a plan you own; we get a client who is choosing the next step with full information instead of on a pitch. If you want the free, faster warm-up first, the AI-Visibility Teardown shows you the gap before you commit — and when it's obvious, the audit is the next step.