How we measure what AI is saying about your organisation

Ask ChatGPT, Gemini, or Google’s AI Overviews to explain your organisation, and yes, you’ll get an answer. But ask again, you’ll get a different one.

That’s the number one challenge when measuring any organisation’s visibility in AI search. Your organisation’s presence in AI answers can be observed, but it’s not necessarily consistent. 

AI doesn’t behave like traditional search. Search results have always varied by location, device, personalisation, and timing, but marketers have long had tools that can aggregate rankings and trends with reasonable confidence. AI is harder to pin down. There isn’t one stable ranking, one fixed result, or one definitive answer that represents what it says about you. Instead, AI answers change depending on the platform, prompt wording, sources used, search location, and context given by the user.

Organisations need to measure AI visibility in the right way: by looking for patterns across a structured set of relevant prompts and painting a clear picture based on the data, not by relying on individual responses that amount to anecdotal evidence.

In this article, I’ll explain how we measure AI visibility for our clients at exceptional™.

Request an AI Performance Analysis to understand how your organisation appears across AI-driven discovery.

You need to understand if and how AI talks about your organisation

AI-generated answers now sit between your audience and your website.

Google’s latest figures say AI Overviews now reach more than 2.5 billion people per month. 

Ofcom has reported that 75% of online adults read AI-generated summaries at the top of search results. 

Pew Research Center found that users were significantly less likely to click a traditional Google result when an AI summary appeared.

Your website still plays a vital role, but AI is increasingly shaping the first impression before someone ever clicks through. To ensure that your organisation is visible and chosen where your audience is making decisions, you need to know whether you’re appearing in AI answers, and how you’re being represented.

You need to look at your AI visibility across the questions buyers actually ask

It can be tempting to ask ChatGPT, Gemini, or AI overviews about your organisation and use its answer to judge how visible and well-represented you are. 

That’s not enough. Your AI visibility can only be accurately understood by looking at a broad data set, which means looking at how your organisation performs across a wide set of prompts. Google’s Liz Reid has pointed out that AI Overviews are leading people to search more and ask new questions that are often longer and more complex, so organisations need to test much more than one obvious query.

You need a large data set to see the big picture patterns: where you appear, where you’re missing, how you’re described, and when competitors are recommended ahead of you.

That means testing prompts across different types of buyer intent. A branded prompt might describe you accurately. A comparison prompt might leave you out. A broad category prompt might mention you. A commercial prompt might recommend competitors instead.

At a basic level, you need to be looking at how AI represents your organisation across a prompt set that covers:

• Your organisation: e.g. “What does [organisation] do?”
• Your sector:  e.g. “Who are the leading organisations in [sector]?”
• Your services:  e.g. “Which companies provide [service] for [audience]?”
• Your competitors: e.g. “How does [organisation] compare with [competitor]?”
• Specific buyer problems: e.g. “Which companies can help with [specific problem]?”
• Provider comparisons: e.g. “What are the best providers for [service]?”

Those examples are far from exhaustive. The greater variety of prompts you’re able to test, the more accurate your understanding of your AI visibility will be.

Most established organisations will be described by AI with relative accuracy if you ask about them directly. The stronger test is whether they appear when someone asks questions that could shape a buying decision, such as:

• Who are the leading providers in this market?
• Which organisations are best for this specific service?
• How does one provider compare with another?
• Which companies have credible experience in this area?

That wider prompt set gives you a more reliable view of your AI visibility. It shows whether AI can do more than recognise you. It shows whether it understands you well enough to recommend you above your competition.

You need to explore the sources AI cites when it talks about you

Once you’ve got a good picture of how your organisation is being represented by AI, the next step is understanding where it’s getting its information from. 

When ChatGPT, Gemini, or Google’s AI results answer a question about your organisation, market, or competitors, they often cite supporting sources. To make things complicated, those citations are not a perfect record of how the answer was built. 

AI citations should be treated as clues rather than a definitive record of how an answer was generated. Different AI platforms retrieve and display sources in different ways, but the sources they surface can still provide a useful indication of which websites, articles, profiles, and pages are influencing how your organisation is being understood.

Look at the sources a given AI model cites most often. 

• Are they your own pages, such as service pages, case studies, or reports?
• Are they third-party sources, such as LinkedIn, press coverage, directories, or review sites? 
• Are they current, accurate, and aligned with how you want to be understood?

In our energy sector research, for example, LinkedIn was the most commonly cited third-party source, appearing in at least one URL in 15.3% of AI responses. This indicates not only that many of the organisations analysed could improve the clarity of their websites, but it also presents a potential opportunity for them to shape the conversation with their owned LinkedIn content. 

By understanding where AI is going to understand your organisation, you can pin-point the areas you need to target to improve how you are being represented in answers. 

How we measure AI visibility in the AI Performance Analyser

Measuring AI visibility properly means testing the questions buyers actually ask, across a large enough prompt set to see patterns rather than one-off results.

That takes more than a few manual searches. You need the right prompts, enough data, competitor comparisons, source analysis, and a clear way to interpret what the results mean.

Our AI Performance Analyser does that work for you. It shows where your organisation appears, how it’s described, which sources are cited, and how competitors perform against the same prompts.

More importantly, it shows what to do next: where your content, positioning, authority, technical setup, or wider digital footprint needs to improve.

Request an AI Performance Analysis to understand what AI is saying about your organisation and what you need to improve.

Key takeaways

• AI visibility can’t be understood from one prompt, one platform, or one screenshot.
• To measure AI visibility properly, you need to test structured prompts across different types of buyer intent.
• The useful signals are not just whether you appear, but how you’re described, which sources are cited, where you sit against competitors, and whether you appear for prompts that matter commercially.
• If you want to understand what AI tools are saying about your own organisation, request an AI Performance Analysis.