Practical Experiments in AI Search Visibility
Practical experiments examining how brands, products and websites appear in AI-generated answers. These tests explore recommendation variability, mentions, citations and share of voice across repeated searches.
If a business tracks one carefully chosen AI search prompt, how much does that result tell us about its visibility for the wider customer need? I tested this by taking one commercially specific accountancy scenario and expressing it five different ways. The customer did not change. The location did not change. The services required did … Read more
If a business’s website is cited in an AI-generated answer, does that mean the business is actually being recommended? I tested this by running the same commercially focused accountancy query through Google AI Mode ten times. The relationship turned out to be surprisingly one-sided. 29 of the 30 recommendation positions included the recommended firm’s own … Read more
I asked Google AI Mode a very simple question: “measure ai brand visibility” Then I asked it again. And again. Five times in total, using exactly the same wording. The answers were remarkably consistent. Google AI Mode repeatedly referred to the same new industry framework, recommended many of the same metrics, and suggested broadly similar … Read more
Choosing the right searches is one of the most important parts of AI visibility testing. But there is a difference between creating a long list of prompts and creating a query set that genuinely represents potential customers. You could test: That gives you four prompts. But they are essentially variations of the same broad customer … Read more
Being cited by AI is not the same as being recommended. Our experiments suggest a business can appear as a source without being presented as a solution — or be recommended while another website provides the supporting citation. That means AI visibility may need to be measured across separate signals such as recommendations, citations, mentions and representation accuracy.
AI visibility is not just about whether your business appears in a recommendation. We compared Google AI Mode’s descriptions of several businesses with what those businesses actually publish on their websites. The results showed a mix of accurate claims, reasonable inference, unsupported associations and important omissions — suggesting that businesses should measure not only whether AI mentions them, but how accurately it represents them.
We tested whether simply adding more detail to an AI search changes the recommendations, or whether the type of detail matters more. In our CRM experiment, a commercially meaningful requirement — temporary staffing — changed both the features AI prioritised and the products it recommended, while a lower-relevance detail such as five years in business had a much smaller effect. The results suggest that decision-relevant specificity may matter more than specificity alone.
We tested whether making AI search queries more specific changed the businesses Google AI Mode recommended. Across accountancy and commercial cleaning, specialist firms became much more visible as the customer’s circumstances were described in greater detail. When we inspected those firms afterwards, their websites often closely matched those specific customer needs. It does not prove causation, but the pattern is worth paying attention to.
AI visibility only matters if you are measuring the queries that could realistically bring you customers. In this practical experiment, we tested how accountant recommendations changed as a search became more specific — then examined the websites that repeatedly appeared. The results suggest that understanding when your business should be the right answer may be more valuable than chasing a generic AI share-of-voice score.
I ran 25 broad Google AI Mode searches across five software markets. The results repeatedly favoured small groups of familiar category-default brands—but more specific customer requirements produced very different recommendations.