Your AI visibility score may be hiding a problem | MarTech

Your AI dashboard’s single score is likely hiding two very different problems.

Semrush’s “2026 AI Visibility Index” found that, across 22 categories and four AI platforms, fewer than half of the top 100 mentioned brands also appeared in the matching top 100 citation lists on average in every month measured.

A single “AI visibility” score can hide whether the brand appears in the answer, whether its website supplies the evidence, or whether third-party sites carry most of the weight. Marketers need to know which of those positions they hold before deciding what to fix.

A mention and a citation are different wins

The report breaks down AI visibility into two metrics. A mention means the brand appears in the answer. A citation means a domain appears as a source.

Ford shows how far apart those positions can get.

In the U.S. in July, Ford appeared inside all four top-100 mention lists. Its domain ranked much lower for citations: 1,055th on ChatGPT, 1,146th on Gemini, 1,320th on AI Mode, and 748th on AI Overviews. The reverse happens too. In U.S. Finance on ChatGPT, Investing.com ranked seventh among cited domains but 50th among mentioned brands.

In the U.K., Tesco and Asda both ranked in the top 100 for mentions on ChatGPT, Gemini, Google AI Mode, and AI Overviews. On ChatGPT, Tesco ranked 25th among cited domains while Asda ranked 606th. On Gemini, their citation positions were much closer: 29th and 60th.

Those are different problems. A brand marketing team for Ford or Tesco would want to know which external pages AI systems cite rather than their own. Investing.com already supplies source material; its weaker position is in the answer itself.

Using the same content tactic for both misses the point.

The platform changes the picture

The measurement problem gets messier because the platforms all behave differently.

Gemini mentioned more distinct brands per answer than ChatGPT in every month Semrush measured in the U.S. ChatGPT, meanwhile, included the most source entries per answer, while Gemini included the fewest.

Their source mixes also differ. In the U.S., news and information sites led ChatGPT citations. AI Overviews leaned more heavily on streaming, business software, online learning, and medical-reference sites. AI Mode concentrated more citation weight among a smaller group of leading domains.

AI Overviews stood apart most sharply. Its leading brand mentions overlapped less with the other three platforms than with any other pairing in every month measured. Its source overlap with ChatGPT was also the lowest.

So a strong result on ChatGPT says little about whether the same brand performs well in AI Overviews.

If customers use several AI search bots, marketers need to know which platform has the gap before spending money to close it.

Third-party sites carry the brand

The old SEO instinct says to start with your website. AI visibility makes that less reliable.

When your brand appears in an AI answer, the cited sources show where that visibility comes from. If those sources are mostly publishers, retailers, community sites, or reference platforms rather than your own domain, the problem may sit outside your website.

In the U.S., Chevrolet, Disney, Ford, Honda, Marvel, Nintendo, and Toyota all ranked in the top 100 for mentions across all four platforms. However, they were all outside the top 500 for citations in every month measured.

That puts some AI visibility work outside SEO.

If the brand appears but its site doesn’t, marketers need to inspect the pages that do get cited. The fix might belong with content, engineering, or communications rather than SEO.

“Create more content” is a weak default when the source gap lies elsewhere.

Category matters as much as platform

Even within the U.S., broad AI visibility benchmarks can mislead.

Semrush found that differences between categories exceeded differences between platforms on mention-to-citation overlap in every month measured.

In Real Estate, for example, the same U.S. source domains appeared across all four platforms every month: Apartments.com, Realtor.com, Redfin, and Zillow. U.S. Industrial had no equivalent recurring source core.

A real estate team can start with a relatively stable set of recurring sources. An industrial team may need a more specific set of competitor and customer questions before a broad-category benchmark tells them much.

A useful AI visibility question needs the customer question, competitor set, platform, and category attached to it.

Start with separate mention and citation baselines for a fixed set of customer questions, then inspect the answers and cited pages before funding a tactic. Keep business outcomes separate from visibility measures, too. Semrush’s study does not establish that a change in AI visibility caused a commercial result.

The useful operating principle is simple: diagnose the type of visibility gap first, then give the work to the team that can actually fix it.

The full report can be downloaded here. (Registration required)