AI Visibility Audit: What It Checks and How to Measure It

The audit works from a fixed set of buyer questions --- the questions your customers would ask an engine about your category --- and records what each engine answers about you. It records whether the machine names your brand, how it describes you, whether the description is accurate, and whether competitors are named instead. The same questions run on the same schedule turn anecdotes into a trend line, which is the whole point: a single snapshot cannot show whether you are improving.

A complete audit checks four things. Share of voice: how often your brand is named versus competitors across the prompt set. Accuracy: whether the machine's description of your business is correct. Sentiment: whether the mention is positive, neutral or negative. Citations: whether answers link to your pages or to third parties. The output is a gap list --- the questions where you should be named and are not --- and that list is the work queue.

Score every answer on a four-point scale: 0 if your brand is absent, 1 if the concept appears without your name, 2 if your brand is named, 3 if it is named and linked. Run the same prompt set across at least two engines, on a schedule, and record the scores. The trend matters more than any single number. Neural Command LLC uses exactly this scoring in its public trials, where every seed carries a marker so citation, mutation or silence can be traced.

Two kinds. Monitoring suites track brand mentions across AI engines on a subscription, and manual workflows query the engines directly with a fixed prompt set and a scoring rubric. The manual method is cheaper and reproducible, and it forces you to own the question list, which is where most audits go wrong: a badly chosen prompt set measures nothing. The first pass can be automated --- Neural Command's free diagnostic runs it against your host.

Monthly is the practical minimum. Engines change, competitor content changes, and a monthly trend catches both. Weekly is justified while you are actively shipping changes, quarterly is enough for a stable brand in a slow category. One rule is non-negotiable: each run uses the same prompt set, or the trend is meaningless.