Semrush Analyzed 126 Million AI Prompts. Most Brands Still Cannot Measure What It Found.
Semrush analyzed 126 million AI search prompts and found that 45% of marketing leaders cannot measure brand visibility in AI answers. Only 9% have the tools. Here is what founders should do about the measurement gap before spending another dollar on AI optimization.
Semrush just released the largest AI search study ever conducted: 126 million U.S. AI search prompts analyzed across ChatGPT, Gemini, Google AI Mode, and AI Overviews. The headline finding should stop every founder cold. 45% of marketing leaders cannot accurately measure their brand's visibility in AI-generated answers. Only 9% have the tools to track it across platforms. The industry is pouring money into AI optimization while nearly half of the people writing the checks have no idea whether it is working.
I have been building measurement infrastructure for this exact problem for years. What Semrush confirmed at 126-million-prompt scale is something I have watched play out one client at a time: the crisis is not that brands are invisible to AI. The crisis is that they cannot tell.
What 126 Million Prompts Reveal About How AI Engines Actually See Your Brand
The Semrush data exposes a structural problem most marketers have not confronted. Each AI engine builds brand perception through fundamentally different citation patterns.
ChatGPT cites an average of 15 sources per response, drawing heavily from community platforms like Reddit and reference sites like Wikipedia. Gemini cites an average of 3, pulling from a narrower pool that includes Wikipedia, Reddit, and YouTube. A brand that performs well in one AI environment can be functionally invisible in another.
This is not a minor variance. It means a single "AI visibility score" is meaningless. A brand that dominates ChatGPT's citation graph because it has strong Reddit community presence might not exist in Gemini's world at all. The reverse is equally true.
Meanwhile, the traffic stakes keep escalating. Adobe's data shows AI traffic to U.S. retail sites surged 1,324% between October 2024 and May 2026. Travel is up 2,215% in the same period. These are not theoretical query volumes. They are real buyers arriving through AI-mediated discovery, and most brands cannot tell where they came from or why they were recommended.
Why Your Current Dashboard Cannot See This
Every marketing team has a dashboard. Google Analytics. Search Console. An SEO tool. Those dashboards track impressions, clicks, rankings, and backlinks. They measure performance on Google's traditional SERP.
None of them tell you what happens when a buyer asks ChatGPT "What is the best project management tool for remote teams?" or when Perplexity processes "Which CRM should a 50-person startup use?"
Searchless.ai tracked 500 brands across three major AI engines over 90 days and found that 88% are never mentioned when users ask for recommendations in their category. Not poorly positioned. Not mentioned at all. And most of those companies are still optimizing for a search environment that is no longer the primary discovery layer for a growing share of their buyers.
The emarketer/Partnerize research frames this as the collapse of click-based measurement. Andy Crossen, Partnerize's Chief Product Officer, put it directly: "SEO is not about ranking pages anymore. It is about fighting to become that trusted, cited source." When AI compresses the purchase journey from ten clicks to one answer, the attribution models that assume a click connects discovery to conversion simply break.
The industry's response so far has been to throw money at the problem. 94% of marketers plan to increase GEO investments this year, according to Conductor. But increasing spend on a strategy you cannot measure is how you turn a measurement gap into a financial hole.
The Earned Media Lever Most Brands Keep Underpricing
Here is where the data converges on something most founders have not internalized.
Muck Rack analyzed 25 million cited links across ChatGPT, Claude, and Gemini in May 2026 and found that earned media accounts for 84% of all AI citations. Journalism alone represents 27%. According to Seer Interactive, the gap between brands with active third-party trust signals and those without is 75x in AI citation frequency. Not 75%. 75 times.
That number reframes the entire AI visibility conversation. Most of the advice circulating right now centers on technical fixes: structured data, schema markup, FAQ content that mirrors how AI engines parse queries. Those are not wrong. They are incomplete. AI does not rank pages. It cites sources it has already decided are credible, and that credibility decision was made through months of editorial coverage, third-party mentions, and earned authority before the buyer ever typed a query.
Sarah Stahl, VP of Marketing at Lake.com, documented this firsthand: "AI doesn't rank. It cites. And what it cites is what the internet has decided is credible over time, before the traveler ever types the question." Her team made a deliberate choice to build earned credibility instead of chasing paid visibility they could not sustain against larger competitors. The AI engines rewarded the approach.
This is not a content optimization problem. It is an earned authority problem. And the distinction between the two is the difference between a brand that gets cited and one that keeps publishing into the void.
How to Audit Your AI Visibility Before Spending Another Dollar
Stop optimizing until you can measure. Here is the sequence that matters.
Run the multi-engine test. Open ChatGPT, Perplexity, Gemini, and Google AI Mode. Do not search your brand name. Search the queries your buyers actually ask when they are evaluating solutions in your category. Track whether you appear, where you appear in the response, and what the engine says about you. Do this across all four platforms, because Semrush's data proves performance varies dramatically between them.
Map your citation sources. When you do appear, trace where the AI engine pulled its information. Was it your owned content? A press mention? A G2 review? A Reddit thread? This tells you which trust signals are actually driving your AI visibility, and which ones are dead weight.
Audit your earned media footprint against AI citation patterns. The 84% earned media citation rate means your PR program is now your most important AI input channel. If your comms team is measuring press coverage by reach and sentiment without tracking AI citation impact, they are measuring the old game.
Measure per-engine, not in aggregate. A single "AI visibility score" averaged across platforms hides the structural variance. ChatGPT's 15-source citation pattern rewards broad community presence. Gemini's 3-source pattern rewards concentrated authority in a narrow set of high-trust domains. Your strategy has to account for both.
This is what I call Machine Relations: the discipline of measuring and managing how AI engines perceive, trust, and cite your brand. It is not a marketing channel. It is the operating layer underneath every channel that AI now mediates. The Semrush data did not create this reality. It quantified it at a scale that makes ignoring it a choice, not an oversight.
The brands that figure out measurement first will be the ones that actually know whether their AI strategy is working. Everyone else will keep spending, keep publishing, and keep wondering why the dashboard says everything is fine while their buyers are getting recommendations from someone else.
FAQ
What is the Semrush AI Visibility Index?
The Semrush AI Visibility Index is a study analyzing 126 million U.S. AI search prompts across ChatGPT, Gemini, Google AI Mode, and AI Overviews. Released in June 2026, it benchmarks brand visibility across 22 industries and reveals that 45% of marketing leaders cannot measure their brand's presence in AI-generated answers.
Why do brands perform differently across AI search engines?
Each AI engine uses different citation patterns. ChatGPT averages 15 sources per response and favors community platforms like Reddit. Gemini averages 3 sources from a smaller pool including Wikipedia and YouTube. A brand visible in one engine can be invisible in another, which is why per-engine measurement is required.
How does earned media affect AI search visibility?
Earned media accounts for 84% of all AI citations according to Muck Rack's analysis of 25 million cited links. Brands with active third-party trust signals are cited 75 times more frequently than those without, making PR and earned media the most important input layer for AI visibility.