AI Search Visibility

Start with an honest picture

Most businesses have a rough sense of where they rank in Google. Very few know what AI assistants say about them. An AI search visibility audit closes that gap. We put the questions your buyers ask into the AI experiences they use and record what comes back: whether you are mentioned, how you are described, which competitors appear in your place, and which sources the engine relied on to decide.

The audit is often the first piece of work we do with a client, because it turns a vague worry (“are we invisible in ChatGPT?”) into a list of specific findings that can be prioritised and fixed. It is also useful on its own, for teams who need evidence before they ask for budget.

How we run the audit

1. Build the prompt set

Good audits start with good questions. We draw them from your sales calls, support tickets, site search, Search Console queries and the way your customers describe their problems. We group them by intent: early research (“how do I reduce no-shows at a clinic?”), category questions (“best appointment software for clinics”), comparisons (“Brand A vs Brand B”) and brand questions (“is Brand A any good?”). We agree the final set with you before testing. Our guide to AEO prompt research explains the method in more depth.

2. Test across engines, more than once

AI answers vary between runs, accounts and locations. We run each prompt several times on each engine in a clean environment and record every result, so that a finding reflects a pattern rather than a lucky or unlucky answer. Engines usually include ChatGPT, Google AI Overviews and AI Mode, Perplexity, Gemini and Microsoft Copilot.

3. Read the answers, not just the mentions

A mention can still be bad news. We note how you are described, whether facts such as pricing, locations or features are correct, how you are positioned against competitors and what tone the answer takes. Errors are often the most valuable finding, because they point to an outdated or inconsistent source somewhere on the web.

4. Trace the sources

Where engines show citations, we record them. Where they do not, we look for the pages most likely to have shaped the answer. This tells us whether the fix lives on your own site, on a profile you control, or on a third-party page you need to earn a place on.

What you receive

• The prompt set, grouped by intent and buying stage
• Results for every prompt on every engine tested, with the date and conditions of the test
• Your share of mentions against the competitors you name
• A list of factual errors and inconsistent descriptions, with their likely source
• The domains most often cited for your topics
• A fix list ranked by effort and likely impact, written in plain language

We walk you through the findings on a call. You keep the prompt set and the method, so the baseline can be repeated later by us or by your own team.

What an audit cannot tell you

An audit is a snapshot of a moving system. It shows patterns, not a permanent score, and it cannot tell you exactly how many people saw a given answer, because the engines do not publish that data. We say so in the report. For ongoing measurement we combine repeated prompt tests with the signals you can measure directly, such as referral traffic from AI tools in your analytics. Our article on AEO KPIs covers the metrics we use, and our guide to tracking AI traffic in GA4 explains how to set up that part yourself.