Key takeaway: A traditional Google ranking is not a complete measure of how visible a brand is in AI search.
Key Takeaway
A traditional Google ranking is not a complete measure of how visible a brand is in AI search. Semrush defines AI visibility through several observable dimensions: whether a brand is mentioned in an AI-generated answer, whether its site is cited, where it appears relative to other brands, and whether the description is positive, neutral, or negative. In its comparison of 10 randomly selected SaaS queries, only 44.3% of the pages ranking in Google’s traditional top 10 appeared in at least one AI-generated answer. That makes SEO ranking useful context, but not a reliable proxy for presence in ChatGPT, Perplexity, or Google AI Mode.
Which AI search visibility metrics should teams track
Measurement should begin with a defined set of platforms and prompts rather than with a broad look at organic traffic. Prompts can cover different stages of a buying journey, including questions about what a product category is, comparisons between options, and evaluation questions about value, pricing, reviews, or advantages and disadvantages. For each prompt and platform, record whether the brand appears, whether the answer links to the brand’s site, where the brand appears in the answer, and how the brand is described.
A manual tracking sheet is enough to create a baseline. It can include the platform, prompt, date, brand mention, site citation, relative position, and sentiment. The same prompt set should be rerun on a consistent schedule so the team can identify changes over time. A dedicated tool may also report broader measures such as share of voice or visibility trends, but those outputs still represent the selected platforms and prompt set rather than every AI search interaction.
The prompt list should reflect real audience questions. The Semrush guidance points to customers, prospects, sales teams, analytics referrals, Google’s People Also Ask results, and relevant online communities as possible inputs. The purpose is not to assume that every question has equal business value, but to make the measurement set more relevant to the decisions the audience is actually trying to make.
How SEO page experience adds evidence without replacing AI measurement
AI visibility and technical SEO should be measured separately, but technical SEO data can help explain whether the pages associated with a brand provide a usable experience. Search Engine Journal’s technical SEO guide treats Core Web Vitals as user-centered performance measurements and distinguishes field data from lab data. Field data reflects real page loads from users and can reveal patterns across groups of pages. Lab data comes from a controlled environment and is better suited to diagnosing specific implementation problems. One type of data does not replace the other.
A practical review can start with the Core Web Vitals report in Search Console to identify groups of pages that share a problem, then use PageSpeed Insights or Lighthouse to inspect representative URLs. The guide recommends testing multiple examples because one lab test is not a definitive answer. Largest Contentful Paint measures when the largest image or text block in the viewport becomes visible. Common sources of poor LCP include slow server response, render-blocking JavaScript or CSS, slow resource loading, and client-side rendering.
These checks can clarify page-experience conditions around content that may be found through search, but they do not establish that improving a Core Web Vital will cause an AI system to cite a page. They are diagnostic evidence for the site experience, not a substitute for tracking AI answers directly.
How to separate visibility measurement from attribution
A disciplined workflow can establish an AI search baseline first and then review traditional SEO and page experience as separate layers. The first layer asks whether the brand appears in an AI answer. The second records whether the answer cites the site, where the brand appears, and how it is characterized. A third layer checks relevant technical conditions, including the field-data patterns and lab diagnostics described above. This structure prevents AI mentions, traditional rankings, site visits, and conversions from being treated as one interchangeable metric.
If the team observes visits from AI platforms, it can record them as a traffic source and relate them to the platform, date, prompt group, and landing page when that information is available. That is more defensible than assigning every later action to AI search. The supplied evidence supports repeatable measurement of AI visibility and page experience, but it does not prove that a particular Core Web Vitals improvement will produce a specific AI citation, ranking, visit, or revenue outcome.
Method and Limitations
Teams should define the target platforms, prompt list, schedule, and fields before collecting results. During manual reviews, preserve the date and record whether the brand was mentioned, whether a site link was included, where the brand appeared, and how it was described. Because AI outputs can change, comparisons are meaningful only when the prompts and review rules remain reasonably consistent. On the SEO side, combine Search Console field data with Lighthouse or other lab diagnostics, test several representative URLs, and validate template-level fixes with the development team.
This approach measures the selected platforms, prompts, URLs, and observation period. AI visibility is not the same as a traditional ranking, and a mention is not the same as a click, lead, or sale. A page passing a Core Web Vitals assessment also does not prove that an AI system will cite it. The source material does not provide a universal method covering every AI platform, every user prompt, or complete revenue attribution, so business conclusions should state their data range and conditions.