A brand that ChatGPT mentions in an AI answer is the same brand ChatGPT cites as its source only about 35% of the time. On Gemini, that overlap between what gets mentioned and what gets cited falls to 30%. Google's AI Overviews keep the two aligned far more often, at 64%. Those figures come from Semrush's 2026 AI Visibility Index, a study of 126 million U.S. AI search prompts collected between January and April 2026 across ChatGPT, Gemini, Google AI Mode, and Google AI Overviews. The gap between appearing in an answer and being trusted as its source is exactly where most AI Share of Voice tracking still gets the story wrong.
What the 2026 AI Visibility Index actually measured
Semrush built the index from 126 million U.S. prompts run through four AI platforms between January and April 2026, tracking more than 1,200 brands across 22 industries. It is the largest public dataset yet of how often brands surface in AI answers and how often they get named as the source behind them. The dataset grew fast to get there: Semrush was working from 89 million prompts as of September 2025. What it tracks sits close to what the industry calls AI Share of Voice, how often and how prominently a brand shows up inside AI-generated answers, though Semrush expresses it as a 0-to-100 index score rather than a percentage.
Two named examples from the index show how differently that score plays out. Patagonia held an index score of 79 to 80 across all four months, built mostly on third-party outdoor-gear review sites such as Reddit, OutdoorGearLab, and REI rather than Patagonia's own pages. Shopify scored 81, and on Google AI Overviews specifically its mentions and citations were nearly balanced in April 2026: 45,098 mentions against 46,342 citations.
Why a mention and a citation are not the same event
A mention is any time an AI answer names a brand. A citation is the answer picking a specific page as its evidence for a claim. Semrush's data shows the two overlap far less than most brands assume, and the size of the gap changes by platform, from 30% on Gemini up to 64% on Google AI Overviews.
The industry has a name for that shortfall: the Citation Gap, the difference between how often a brand is mentioned and how often it is actually cited as a source. A wide citation gap usually means an AI engine trusts a brand's category enough to name it from memory, but hasn't found (or doesn't trust) the brand's own pages enough to point to them. That distinction carries an accuracy risk too. When an engine mentions a brand without citing anything behind it, it is leaning on training data or a loose match, which is also where factual errors about brands tend to creep in.
A brand's AI Share of Voice score describes how often it gets mentioned. Its citation rate describes something more valuable: whether the AI engine trusts that brand's own content enough to point a reader straight at it.
Citation density is wildly different across engines
ChatGPT answers cite an average of 15.4 sources each. Gemini cites 3.3, a gap of more than four to one. That difference means the two platforms reward different strategies: ChatGPT has room for a long tail of supporting sources, while Gemini's answers only have space to trust a handful.
Google AI Mode sits at 11.4 sources per response and Google AI Overviews at 9.2, both closer to ChatGPT's end of the range. For a brand chasing visibility on Gemini specifically, that scarcity of citation slots is compounded by how Gemini chunks and retrieves passages before it ever compares candidates: fewer slots mean every passage has to win on relevance alone, without the redundancy a 15-source ChatGPT answer allows. The same scarcity shows up on the traffic side. Platforms that cite fewer sources per answer also tend to send less referral traffic back for what they crawl, the pattern behind the crawl-to-refer ratio.
Visibility concentrates hard in some industries, barely at all in others
In News and Media, the top three brands claim 82.9% of the visibility held by the category's top 10, with Reddit alone pulling 25.8 million mentions. In Finance, the top three hold just 41.4%, the most evenly distributed of the 22 industries Semrush tracked.
Consumer Electronics looks like News and Media: its top three brands take 76.9% of top-10 visibility, with Apple's 11.2 million mentions doing most of that work. Industrial sits at 42.2%, close to Finance's spread. The pattern tracks how many distinct, citable sources exist in a category. Finance and Industrial have large numbers of specialized publishers, forums, and comparison sites all competing for citation; News and Media has fewer entities an AI engine would treat as a primary source in the first place. Running a competitive AI Share of Voice analysis inside a concentrated category is a different exercise than in a fragmented one. In Finance, a challenger brand has real room to take share from the leaders. In News and Media, the top three are unlikely to move.
Only 36 brands show up everywhere, and the drop-off past them is steep
Semrush tracked more than 1,200 brands across its 22 industries, and only 36 of them held top-100 visibility on ChatGPT, Gemini, Google AI Mode, and Google AI Overviews every single month from January through April 2026.
Even inside that group of 36, mention volume falls off fast. YouTube led with 47.9 million mentions, and Nintendo, still inside the universal top 36, closed the list at 1.5 million, a roughly 32-fold spread inside what is already the most consistently visible slice of brands Semrush tracked. For a brand outside that group, showing up on more than one platform, month after month, is closer to luck than strategy without a baseline to manage against. A methodology for calculating AI Share of Voice from scratch is the starting point for closing that gap.
What this means for measuring your own AI Share of Voice
Mention rate and citation rate are two separate numbers, and Semrush's data suggests most marketing teams are only tracking one, if either. Only 9% of marketers surveyed alongside the index said they could track mentions, citations, overlap, and consistency together. 45% said they couldn't measure AI visibility accurately at all.
The teams that closed that measurement gap saw a real business difference: 81% of organizations running an integrated SEO and AI-visibility strategy reported more traffic or leads, against 36% of teams that kept the two efforts separate. Practically, that means tracking mention rate and citation rate as different numbers rather than one blended score, and breaking both out by platform rather than averaging them, since a brand that is strong on ChatGPT's 15-source answers can be nearly invisible inside Gemini's three-source ones. A step-by-step measurement approach matters more here than a single benchmark, since the right target differs by platform and by how concentrated a brand's industry already is.
The index has real limits worth naming. It covers ChatGPT, Gemini, Google AI Mode, and Google AI Overviews, drawn entirely from U.S. prompts. Claude and Perplexity sit outside this particular dataset, and non-U.S. query behavior does too, both of which can change a brand's citation mix substantially according to Semrush's own published methodology. A brand whose buyers research heavily on Perplexity, or outside the United States, should treat these numbers as a floor on the full picture, not the whole of it.
Frequently Asked Questions
What is the AI citation gap?
The citation gap is the difference between how often a brand is mentioned in an AI-generated answer and how often it is actually named as the cited source behind a specific claim. Semrush's 2026 data puts that gap at its widest on Gemini, where only 30% of mentioned brands are also cited, and at its narrowest on Google AI Overviews, at 64%.
How many AI prompts did Semrush's 2026 AI Visibility Index analyze?
Semrush analyzed 126 million U.S. AI search prompts collected between January and April 2026, run across ChatGPT, Gemini, Google AI Mode, and Google AI Overviews, tracking more than 1,200 brands across 22 industries.
Why do some industries show more concentrated AI visibility than others?
Concentration tracks how many distinct, citable sources exist in a category. News and Media and Consumer Electronics have a handful of dominant entities and fewer competing publishers, so their top three brands take 82.9% and 76.9% of top-10 visibility. Finance and Industrial have more specialized publishers and comparison sites, spreading visibility more evenly, with the top three holding only 41 to 42%.
Does an AI engine mentioning a brand mean it will also cite that brand's page?
No. Semrush's data found mention and citation overlap only 30% to 64% of the time depending on the platform. An AI engine can name a brand from general knowledge or a competitor's comparison page without ever citing the brand's own site as a source.
How many brands maintain visibility across every major AI platform?
Only 36 of the more than 1,200 brands Semrush tracked held top-100 visibility on ChatGPT, Gemini, Google AI Mode, and Google AI Overviews in every month from January through April 2026, and mention volume inside that group of 36 still varied by more than 30-fold.
What should marketers change based on this data?
Track mention rate and citation rate as separate numbers instead of one blended score, break both out by platform since citation slots range from about 3 per answer on Gemini to more than 15 on ChatGPT, and treat cross-platform consistency, not a single high score, as the real benchmark.