Brands Are Tracking AI Visibility, But Are They Measuring The Right Things? via @sejournal, @MattGSouthern

AI visibility tools now make it easier for you to see how often your brand shows up across ChatGPT, AI Overviews, AI Mode, and other AI search surfaces. While the scores they provide can be helpful for guiding marketers, it’s not always clear what actions to take based on them.

The demand for these tracking tools is steadily increasing. Ahrefs reports a 184% rise in searches for “AI search tracking” and a 175% rise for “AI rank tracking” over the past year in the U.S., based on monthly data from their Keywords Explorer. While these figures indicate growing interest rather than widespread adoption, the trend is clear and unmistakable.

A change in your visibility score, whether it goes up or down, doesn’t automatically mean your strategy is perfect or needs tweaking. To truly understand that score, it’s helpful to know what each metric reveals, and what it might overlook.

All the data here is from Ahrefs, which publishes measurement research on AI search and sells an AI visibility tool called Brand Radar.

A Mention And A Citation Aren’t The Same Thing

An AI visibility score can combine at least two separate measurements. One is a mention, which happens when a brand’s name is included in a response. The other is a citation, which is a URL linked as a source. Sometimes, only one of these appears, and each can have its own challenges. For example, a mention without a citation means the brand is mentioned but not linked, while a citation without a mention makes the page a source while the answer discusses something else. Combining these into a single score can sometimes make it unclear which factor caused the change. Therefore, it’s helpful to consider what specific event led to a score shift whenever it happens.

Retrieval adds an extra layer of complexity. Ahrefs recently examined 1.4 million ChatGPT prompts and found that URLs from Reddit were retrieved at scale but cited in only 1.93% of cases. Some of these URLs might never have been opened or read, so the data doesn’t necessarily show their influence on the final response. Essentially, a citation report highlights which sources are credited, but it doesn’t capture every detail of the retrieval process. We talked about this report back in April.

Keep in mind that tools counting these events differently may lead to disagreements about whether visibility has increased.

Traditional Search Visibility Still Matters

Traditional rankings account for a meaningful share of AI Overview citations without being a requirement for them. In Ahrefs’ analysis of 863,000 keywords and 4 million AI Overview URLs, 37.1% of these URLs also appeared in Google’s organic top 10 for the same query. Meanwhile, 26.2% ranked between 11, and 100, and 36.7% didn’t make it into the top 100.

An earlier version of the analysis showed higher overlap, but Ahrefs noted improvements in their citation parsing, which means the older data isn’t directly comparable. We covered the latest findings in March.

Ahrefs highlights that Google’s query fan-out process helps explain citations from outside the usual rankings. When an AI Overview is generated, Google can break down the query into related sub-queries. This allows pages that do well across related topics to be cited, even if they don’t rank for the original question. Keep in mind, though, that rank tracking focused on main terms might not fully capture this activity.

See also: Google Says Keyword Fragmentation Is Reshaping AI Search

The Signals Associated With AI Visibility Aren’t Necessarily What SEOs Expect

Schema markup is an example how some people might misunderstand visibility data. Ahrefs research shows pages cited by AI were almost three times more likely to include JSON-LD than those that weren’t cited. They tracked 1,885 pages that added JSON-LD between August 2025 and March 2026, compared them to 4,000 control pages, and carefully examined citation patterns across three different platforms.

Interestingly, adding schema didn’t significantly boost citations for any platform. For AI Mode, citations increased by 2.4% compared to controls, and for ChatGPT, by 2.2%, but Ahrefs considered these differences statistically insignificant. Citations for AI Overview fell by 4.6%, a small yet statistically significant decrease. This amounts to about 12 fewer citations per page daily, in a context where most pages received hundreds, and both groups had been declining even before schema was implemented. Ahrefs says it can’t tell whether schema caused the decline or if other factors played a role.

The sample included pages with at least 100 AI Overview citations before adding schema, focusing on pages already referenced by AI systems. However, it doesn’t indicate if schema assists a page in gaining initial recognition, which we covered in May.

The same ChatGPT analysis found a correlation schema didn’t show. Cited pages’ titles aligned more closely with the sub-queries ChatGPT generates from a prompt than non-cited titles did, and the gap was wider against those sub-queries than against the original prompt. Ahrefs noted that URLs with natural-language slugs were more frequently cited. They used open-source tools to estimate relevance, as ChatGPT’s internal processes are proprietary. Overall, these results specifically relate to ChatGPT, and may not reflect broader AI search patterns.

See also: Schema, LLMs & The Low Bar For ‘Evidence’ In GEO

What Do You Do With The Data?

Before making any decisions based on visibility metrics, ensure the measurement is consistent. Ahrefs highlights that AI responses are probabilistic, meaning a prompt can produce different brands, and citations each time. They recommend analyzing large prompt sets to identify overall patterns. Sometimes, score changes happen because of updates to prompts or switching models, which indicate measurement variability rather than real market changes.

Once you’ve verified that your measurements are reliable, each metric can offer valuable insights into specific questions.

For example, if mentions decrease while prompts remain steady, it might be helpful to look into whether competitors have gained the mention share you lost, identify which topics are seeing declines, and see if these changes are happening on just one platform. Asking these questions can really help you figure out whether you’re facing a positioning challenge or just a platform-specific glitch.

If citations decrease but rankings stay steady, look at which pages are being cited instead, and whether your current coverage matches how the queries decompose. The fan-out pattern above makes this worth checking, even though data by itself can’t pinpoint the exact reason.

When rankings drop, that points back to traditional search performance, and the overlap numbers above are one reason those changes belong in an AI visibility review.

Overall, the data guides us in deciding which question to explore first, rather than giving us an instant diagnosis.

Stop Treating AI Visibility As A Score

A visibility score can go up even if clicks decrease. Ahrefs looked at clickthrough rates for 300,000 keywords using combined desktop Google Search Console data, comparing December 2023 to December 2025. They found that for keywords with an AI Overview, the top result’s CTR was 58% lower than expected without it. Keywords without AI Overviews also saw CTR drop during this time, so the 58% describes an added decline beyond the existing trend. It’s a correlation drawn from informational queries, not a sitewide traffic measurement.

A brand can maintain its rankings and even earn AI citations, but might still see fewer clicks because being featured in an answer and driving traffic are two different things.

Combining visibility data with actual performance results can really help close this gap. Search Console’s AI performance reports are a great, free way to get started. They show impressions for AI features, though the dedicated report doesn’t include click or query data as of publication. Adding analytics on AI referrals can give you even deeper insights, making the visibility score just one of several important indicators rather than the whole story.

When tracked this way, the score reflects how frequently a brand shows up in prompts that the tool watches, highlighting areas that might need a closer look, and pointing out where there may be gaps. Remember, the true success of a strategy really depends on the real outcome data we gather.

To dig more into this theme, join us for the SEJ webinar “AI Cites Your Brand. Now What? Turn AI Visibility Data Into Actions”  sponsored by Ahrefs on September 15 at 12 p.m. ET. The event will focus on key metrics and how to prioritize next steps.

More Resources:

Featured Image: Rawpixel.com/Shutterstock