Why this page exists
Generative Engine Optimisation is a young field. There’s a lot of noise. Most publishable research lives behind gated PDFs, in conference talks, or buried in customer-testimonial blog posts.
We read this stuff so you don’t have to. We pull out the stats that matter, attribute them clearly, link to the deep-dive, and update the page when new research lands.
Stats are nothing without judgement. The numbers below tell you which way the field is moving. They don’t tell you which ones matter for
your business — that depends on your sector, your buyers, and the specific decisions those buyers are making in front of an AI. That’s the work
AVS is built to do.
AI Search Visibility — the headline numbers
The freshness, signal-architecture and Binary-Cliff figures the field has converged on.
70% of AI-cited pages were updated within the past year. Freshness is a primary citation signal.
Content less than three months old is 3× more likely to be cited.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
60%+ of commercial pages cited by ChatGPT were updated in the past six months.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
Pages going more than three months without an update are 3× more likely to lose AI visibility.
AirOps, The 2026 State of AI Search (companion report with Kevin Indig), February 2026.
Brands strong on entity authority, third-party validation and community discussion appeared 7.8× more often in AI answers than brands weak on those layers.
The Binary Cliff is real. Brands are either in the consideration set or invisible — real-world performance does not carry over to AI surfacing on its own.
Seer Interactive, GEO Olympics Study, February 2026.
Sources & third-party citations
Where AI is actually pulling brand mentions from — and why owned content alone won’t cut it.
85% of brand mentions in AI search come from third-party sources, not the brand’s own site.
AirOps, Complete AI Search Playbook for Marketers, March 2026 (third-party analysis across 21,000+ brands).
Brands are 6.5× more likely to be cited via third-party sources than from their own domain.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
~90% of third-party citations come from listicles, comparison pages and review sites.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
80% of cited brands appear in the first three positions of those listicles, comparisons and reviews.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
96% of AI citations come from earned media. Press releases account for just 0.04%.
95%+ of AI citations come from non-paid media. 27%+ are from journalism.
Muck Rack, citation analysis (1M+ AI citations).
Wikipedia is the single most-cited source across the GEO Olympics dataset.
Seer Interactive, GEO Olympics Study, February 2026.
Schema, structure and on-page signals
The on-page elements that correlate with AI citation. Structural numbers are the most rigorous bit of the AirOps dataset (12,000-page analysis).
Pages with three or more schema types are 13% more likely to earn AI citations.
AirOps, Complete AI Search Playbook for Marketers, March 2026 (12,000-page structural analysis).
Pages with FAQs have a 40% higher likelihood of being cited.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
Clear H1 → H2 → H3 hierarchy gives 2.8× higher citation odds.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
Lists or tables appear in nearly 80% of ChatGPT citations vs 29% in Google’s top results.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
Adding citations, quotations from authoritative sources and statistics to content can improve LLM citation rates by up to 40%.
Princeton / Georgia Tech academic GEO research, 2024.
Every structural element tested — FAQs, schema, lists, headings — appeared more frequently in ChatGPT-cited content than in Google’s top results, often by margins of 20–40 percentage points.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
Multi-model variation
Why measuring one LLM gets you about a third of the picture.
68% of brand mentions in AI search appear in only one model. Cross-model consistency is rare.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
ChatGPT, Perplexity, Gemini, Claude and Bing Copilot do not cite the same businesses for the same queries. A business strong on one platform can be invisible on another.
K&C audit observations · also noted by Muck Rack, Authoritas, and Georgia Tech academic GEO research (2024).
Roughly 1 in 5 LLM mentions of a brand are framed in stale or out-of-date language. The description AI returns of you may not be the description you want.
Seer Interactive, GEO Olympics Study, February 2026.
Methodology disclosure & sample sizes
How big these datasets actually are — and how many prompts per brand they translate to.
~15 million data points were analysed in the AirOps 2026 dataset across AI answers, queries, citations and brand mentions — across 21,000 brands. Per-brand average: ~700 prompts.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
231,347 LLM responses across six LLMs were measured in the Seer GEO Olympics Study, mapped to every category in the 2026 Winter Olympics.
Seer Interactive, GEO Olympics Study, February 2026.
5.1 million AI responses, 57.2 million citations were analysed in the AirOps / Kevin Indig State of AI Search companion dataset (60-day window, December 2025 – February 2026, 50 brands, 7 verticals).
AirOps, The 2026 State of AI Search, February 2026.
6,000+ prompts per brand in a full AVS Annual measurement — every prompt designed for the brand’s sector, buyers and audience. Roughly 8× the per-brand depth of the AirOps 2026 average.
Buyer behaviour and AI search adoption
What buyers and search platforms are actually doing — the demand-side numbers.
B2B buyers are building shortlists in AI tools before they visit a vendor’s website.
50% of Google searches already surface AI summaries. Projected to reach 75% by 2028.
McKinsey, cited in AirOps, Complete AI Search Playbook for Marketers, March 2026.
93% of Google AI Mode searches end without a click to any external website.
ChatGPT processes 1.6 billion daily search queries but sends 190× less traffic to websites than Google.
Daily Mail organic CTR fell from 13% to 5% on desktop, 20% to 7% on mobile when Google AI Overviews appear above search results — an 80–90% drop.
Community signals — Reddit, YouTube, LinkedIn
Where community-cited sources sit in AI answers, and which categories rely on them.
48% of AI citations come from Reddit, LinkedIn and YouTube combined.
AirOps, Complete AI Search Playbook for Marketers, March 2026 (5.5M-answer analysis).
Reddit appears as a cited source in ~22% of AI-generated answers.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
75% of YouTube citations occur in non-branded “how-to” queries — instructional content, not brand-search.
AirOps, Complete AI Search Playbook for Marketers, March 2026.
Reddit’s weight as an AI citation signal is heavily category-dependent. Strong in some sectors; almost absent in others.
Seer Interactive, GEO Olympics Study, February 2026.
Four cluster types of community-cited platforms identified: Community Q&A (Reddit, YouTube), Social (LinkedIn, X), Community editorial (Wikipedia, Medium), Review/rating (G2, Trustpilot).
AirOps, Complete AI Search Playbook for Marketers, March 2026.