AI Visibility Audit

An audit is a documented evaluation of how a business currently appears across AI-powered search and answer platforms. It establishes a factual baseline: what AI systems say about the business today, which sources they draw on, and where the record is wrong, incomplete, or inconsistent.

What the audit evaluates

Citations. Whether AI-generated answers link to or reference the business’s own website and profiles — and, when they do, which pages are used and for which kinds of questions.

Brand mentions. Where the business is named in an answer without any citation attached. A mention and a citation are different things, and the audit records them separately, because they mean different things for how discoverable and referenceable a business actually is.

Source references. The sources AI systems use when discussing the business, whether or not the business controls them. In many cases an AI answer about a company is assembled from directories, review platforms, news coverage, and other third-party pages rather than the company’s own website. The audit identifies which sources are doing that work.

Accuracy. What AI systems state about the business — its products, services, locations, history, and leadership — checked against verified facts. Errors, outdated information, and claims that appear nowhere in the documented record are flagged individually, with the answer text that contains them.

Entity consistency. Whether the business’s core information is consistent across the sources AI systems draw on. Conflicting names, addresses, descriptions, and product details across the web give retrieval systems an unclear picture of what the business is, and the audit documents where those conflicts exist.

Competitive presence. Which competing businesses appear in the same category-level answers, and what sources support their presence.

How the audit is conducted

The audit follows the research and documentation process described in our methodology. A fixed set of questions is agreed in advance — the questions customers actually ask about the business, its products and services, and its category. Each question is then run on the relevant AI platforms, and every run is recorded with the platform, the exact prompt, the date, the full response, and the citation URLs the response contained.

Because AI answers vary between runs, key questions are tested more than once, and the report distinguishes between what was consistently observed, what appeared once, and what could not be established. Documented behavior, independent observation, and inference are labelled as such rather than blended together. Findings about accuracy are checked against primary sources and official company documentation wherever those are accessible.

Platforms are reported separately

The audit does not merge its findings into a single score, because the platforms do not work the same way. ChatGPT, Perplexity, Google AI Overviews, and Gemini each draw on different sources, retrieve information differently, and present citations in different forms — and each changes its behavior on its own schedule. Findings are therefore reported platform by platform, so that a strength on one system is not used to paper over an absence on another, and so that later measurements can be compared like for like. Where a platform’s mechanism is unpublished, the report says what was observed and leaves the mechanism alone.

When a business needs an audit

An audit is the right starting point whenever decisions are about to be made on the basis of AI visibility — before commissioning development work, before acting on a suspicion that AI answers contain errors, and before accepting anyone’s claim, reassuring or alarming, about how a business appears in AI-generated answers. In each case the audit replaces the claim with a dated record of what the platforms actually returned.

What you receive

You receive a written findings report covering the question set and how it was tested, the per-platform findings, the sources AI systems relied on, each accuracy issue identified with its supporting evidence, and the entity consistency problems found across the business’s wider digital footprint. The report states plainly what the evidence does and does not show.

What the audit is not

An audit is not a ranking report, and it does not predict how any platform will behave in the future. It is a snapshot of documented observations at the time of testing; AI answers change, and a later test may produce different results. The audit is also not an endorsement. Findings are reported as they are found — accurate, inaccurate, favorable, or unfavorable.

An audit carries no guarantee of any outcome. It cannot promise citations, recommendations, rankings, or visibility, and it does not claim that any later change in an AI answer was caused by any particular action. What it provides is the documented starting point that every other decision depends on: it is the natural baseline for AI Citation Development work and for ongoing AI Visibility Monitoring.

The terms of the engagement

The AI Visibility Audit is a paid service. Payment covers the research, testing, and documentation described on this page. It does not purchase favorable findings or editorial conclusions of any kind, and it does not purchase citations, rankings, recommendations, or visibility on any platform. Our editorial standards govern how findings are reported, for paying clients exactly as for everyone else.