Facts checked 13 October 2026. BlogAwesome isn't affiliated with Semrush, OpenAI, Google or any brand named here.
Written by the BlogAwesome team – we build the tool that checks whether AI names your brand, so we read these studies closely.
Semrush published its 2026 AI Visibility Index on 13 October 2026. The most useful finding for anyone who markets a brand is simple: AI naming your brand and AI citing your website are two different kinds of visibility, and they often don't line up. Here's what the study found, and how founders, agencies and in-house teams can measure and fix both this week.
What did Semrush's new AI Visibility Index find?
The index covers January to July 2026. Semrush ran fixed monthly sets of 2.5 million UK prompts and 25 million US prompts across ChatGPT, Gemini and Google AI Mode, plus a separate Google AI Overviews dataset.
The headline: every month, in both markets, fewer than half of the 100 most-mentioned brands also appeared among the 100 most-cited domains. In the UK, the overlap ran between 36.7 and 45.3 out of 100. In other words, AI often talks about a brand while pulling its facts from somebody else's website.
If you're new to the idea, start with what AI brand visibility is and why it matters.
What's the difference between an AI mention and an AI citation?
A mention is when AI names your brand in its answer. A citation is when AI uses your website as a source for that answer.
Semrush's UK supermarket example shows the gap. Tesco and Asda were both in the top 100 mentioned brands on every platform, every month. But in July, Tesco was ChatGPT's 25th most-cited UK domain, while Asda was 606th. On Gemini they were 29th and 60th. Both brands get named. Only one is consistently treated as a source.
Why that matters for a smaller brand:
Mentioned but not cited: AI is describing you from other people's pages – reviews, roundups, old articles. You don't control what it says, and it can be wrong or out of date.
Cited but not mentioned: your page supplies the facts, but the answer recommends someone else. You do the work; a competitor gets the credit.
Why doesn't doing well on ChatGPT mean doing well everywhere?
Because the engines don't agree. In Semrush's UK data, only 16.7% to 26% of top domains showed up across all four AI surfaces. A brand that looks strong in ChatGPT can be close to invisible in Gemini or Google's AI answers for the same question.
So a single-platform check gives you a false sense of where you stand. We walked through a hands-on version of this yesterday in Does ChatGPT Recommend Your Startup? How to Check (and Fix It).
Who should own AI brand visibility – SEO, PR or brand?
Semrush's answer is that it needs shared ownership, and its data backs that up. News sites made up 29.5% of citations among ChatGPT's top 100 cited UK domains, against 19.6% in the US. Earned media – press, PR, coverage – is a big part of what AI reads.
In practice:
Founders: you own it. Nobody else is going to check what AI says about you.
Agencies: add mentions and citations to client reporting, next to rankings and traffic.
In-house teams: make it one shared metric across SEO, PR and content, rather than three separate reports. Our post on generative engine optimization for brand teams goes deeper.
How do you measure mentions and citations separately this week?
Write down 10–20 questions your buyers actually ask, like "best tool for X" or "who helps with Y".
Ask the same questions in ChatGPT, Gemini, Perplexity and Claude.
For each answer, log: were you named (yes/no), in what position, which URLs were cited, whether your own site was cited, and who was named instead.
Keep mentions and citations as two columns, not one blended score.
Re-run it weekly with the same questions so you can see movement.
For a fuller method, see the AI visibility tracking playbook and how to build a keyword-level AI recommendation audit in one afternoon.
What do you fix when you're mentioned but not cited (or cited but not mentioned)?
What you see | What it usually means | What to do |
|---|---|---|
Mentioned, not cited | AI knows you from other people's pages | Publish clear, answer-first, sourced pages on your own site for those questions, and use the same name and category description everywhere |
Cited, not mentioned | Your page is useful, but AI doesn't connect it to a recommendation | Sharpen your one-line category description, add honest comparison pages, and earn third-party mentions in press, roundups and communities |
Neither | AI doesn't see you for that question yet | Start with a list of the gaps – here's how to map them – and work through them one post at a time |
This is the weekly loop BlogAwesome is built around: check whether ChatGPT, Gemini, Perplexity and Claude recommend you for your keywords, see the gaps and who AI names instead, then research, draft and publish posts in your voice to your CMS – with you approving each one.
A note from us
We're a small startup, and we're still climbing on Google for most of our tracked keywords. We're measuring and fixing it in public, one checked, sourced post at a time – this one included, written with BlogAwesome.
If you want to see where your brand stands, get a free AI visibility audit.