Are Brands Tracking the Right AI Metrics?

AI visibility tools are giving marketers new ways to monitor how often their brands appear in ChatGPT, Google AI Overviews, AI Mode and other AI-powered search platforms.

However, a visibility score on its own does not necessarily explain whether a brand’s strategy is working. Understanding the different measurements behind that score can help marketers work out whether a change represents a genuine improvement, a problem with visibility or simply a change in how the data is being collected.

Interest in these tools is also increasing. Ahrefs reports that searches for “AI search tracking” in the US have risen by 184% over the past year, while searches for “AI rank tracking” have increased by 175%.

The growing interest makes it increasingly important for marketers to understand what AI visibility figures actually tell them.

Mentions and Citations Measure Different Things

One of the first distinctions to make is between a brand mention and a citation.

A mention occurs when an AI-generated response names a brand, while a citation refers to a webpage that the system links to as a source.

These two events do not always happen together. An AI system could mention a company without linking to its website, or cite a webpage without directly mentioning the brand in its response.

If both measurements are combined into one visibility score, it can become difficult to understand exactly why that score has changed.

There is another complication: being retrieved by an AI system does not necessarily mean a source will eventually appear as a citation.

Research from Ahrefs examining 1.4 million ChatGPT prompts found that Reddit URLs were retrieved frequently but were cited in only 1.93% of cases. This suggests that retrieval and citation should be treated as separate parts of the process.

Traditional Rankings Still Have a Role

AI search does not completely replace conventional SEO.

Ahrefs analysed 863,000 keywords and around four million URLs appearing in Google AI Overviews. It found that 37.1% of cited URLs also ranked within Google’s traditional top 10 for the same search.

A further 26.2% ranked between positions 11 and 100, while 36.7% did not appear within the top 100 organic results.

This shows that strong traditional rankings can contribute to AI visibility, but they are not essential in every case.

Google’s query fan-out process helps explain why. An AI system can break a user’s original search into several related queries and retrieve information from pages that perform well for those supporting topics.

As a result, monitoring only a brand’s main keywords may fail to capture some of the pages being used to generate AI responses.

Don’t Assume Schema Automatically Improves AI Visibility

Schema markup is another area where AI visibility data can be easy to misinterpret.

Ahrefs found that webpages cited by AI systems were almost three times more likely to contain JSON-LD than pages that were not cited. At first glance, that might suggest a connection between structured data and AI citations.

However, a separate test did not show that adding schema directly produced a meaningful increase in citations.

Ahrefs studied 1,885 pages that added JSON-LD between August 2025 and March 2026 and compared them with 4,000 control pages.

For AI Mode, citations increased by 2.4% compared with the control group, while ChatGPT citations increased by 2.2%. Ahrefs considered both differences statistically insignificant.

AI Overview citations actually fell by 4.6%, although this was statistically significant. The difference amounted to roughly 12 fewer citations per page each day, while the pages involved were already receiving hundreds of citations.

The research therefore does not establish that adding schema improves AI visibility. Other factors could have influenced the results, and both groups were already experiencing declining citation numbers before the schema was introduced.

Relevance May Matter More Than Expected

The same research uncovered another potentially useful signal.

For ChatGPT, pages that were cited tended to have titles that matched the related sub-queries generated from the original prompt more closely than pages that were not cited.

This difference was more noticeable when comparing pages against the sub-queries rather than the original user prompt.

Ahrefs also found that URLs using natural-language structures were more commonly cited in its ChatGPT analysis.

These findings suggest that understanding the subjects and related questions surrounding a search may be more useful than focusing exclusively on the exact wording of the original keyword.

However, the results relate specifically to ChatGPT and should not automatically be applied to every AI search platform.

What Should Marketers Do When Visibility Changes?

The first step is to establish whether a change in an AI visibility score is actually meaningful.

AI responses are not completely fixed. The same prompt can produce different brands, sources and answers depending on the circumstances, including which model is being used.

For this reason, analysing a large collection of prompts is more useful than relying on individual searches.

Marketers should also check whether the tracking system has changed its prompts, models or methodology. A sudden movement in the score could reflect a measurement change rather than a genuine shift in brand visibility.

Once the data has been validated, individual metrics can help identify where further investigation is needed.

For example, if brand mentions fall while the tracked prompts remain consistent, marketers could investigate whether competitors are appearing more frequently, whether the decline is concentrated around particular topics or whether it only affects one AI platform.

If citations decline while conventional rankings remain stable, it may be worth examining which competing pages are now being cited and whether the brand’s content adequately covers the related topics generated by AI systems.

Visibility Does Not Automatically Mean More Traffic

A brand can become more visible in AI search without receiving more visitors.

Ahrefs analysed click-through rates for 300,000 keywords using combined desktop Google Search Console data from December 2023 and December 2025.

For searches that displayed an AI Overview, the top organic result received a CTR that was 58% lower than expected compared with searches without an AI Overview.

This figure represents an additional decline on top of broader changes in click-through rates during the period, and the research focused on informational searches rather than measuring an entire website’s traffic.

The important point is that being visible in an AI answer and generating a website visit are not the same outcome.

Look Beyond the Visibility Score

AI visibility should therefore be treated as one measurement within a broader performance picture.

Google Search Console’s AI performance reporting can provide useful information about impressions generated through AI features, while analytics data can help businesses identify traffic coming from AI platforms.

Combining these sources can give marketers a clearer understanding of whether increased visibility is translating into meaningful results.

A visibility score can show how frequently a brand appears within the prompts being monitored, but it cannot by itself tell a business whether that exposure is improving leads, sales, traffic or customer engagement.

Measure the Outcome, Not Just the Exposure

The growing number of AI search tracking tools makes it tempting to treat visibility as the main goal.

However, a higher score is not automatically a sign of success, just as a lower score does not necessarily mean a strategy has failed.

The more useful approach is to understand why the score changed and then connect that information to actual business results.

Mentions, citations, rankings and traffic each answer different questions. Used together, they can help marketers identify genuine opportunities and avoid making major strategic decisions based on a single AI visibility number.

As AI search continues to develop, brands will need to move beyond simply asking whether they appear in AI answers and start measuring whether that visibility is accurate, useful and producing meaningful results.