Press kit
The 2026 AI Citation Visibility Study: how 50 crypto protocols appear in AI-generated answers
Original research auditing AI citation behaviour across ChatGPT, Perplexity, and Google AI Overviews. Published 10 May 2026, updated to version 2 on 17 October 2026. DOI-assigned. Full dataset available.
Headline findings
Six findings worth covering
Each can be checked in the public dataset. Version 2 includes a clean citation list and a recount tab with every figure below.
Third parties supplied 92% of Google AI Overview citations and 91% of Perplexity citations. Official protocol pages made up 7.7% and 8.8%. On Perplexity, every official citation went to one protocol, Chainlink.
YouTube received 94 Google AI Overview citations. Reddit received 71. Each beat the combined total for every official protocol page on both platforms (52).
On risk and security questions, official pages made up 1.2% of citations. On comparison questions, 2.5%. Protocols are least visible on their own sites at the moment users are deciding between them. Even when a protocol was named in the prompt, 83% of Google AI Overview citations went to third parties.
Six protocols had zero AI presence across all three platforms: Pendle, Ethena, Sui, The Graph, Filecoin, and Arweave.
Schema markup didn't predict AI visibility. Pendle and Filecoin both have JSON-LD schema and appeared in zero AI answers. Aave has no schema and ranks 2nd. An Ahrefs study published one day later reached the same conclusion.
Ethereum, the highest-scoring protocol, scored 81.9 out of 100, 18 points clear of Aave. The median score was 29. Only 8 of the 50 protocols scored above 50. Outside a handful of long-established names, low AI visibility is the norm.
Study at a glance
Scope and methodology summary
Each protocol was evaluated against a 100-point framework covering five dimensions: AI answer presence across platforms, citation quality, website retrieval readiness, schema and entity clarity, and documentation quality. Protocols were selected using publicly available rankings from DefiLlama, L2Beat, and CoinGecko to represent a range of sizes, ages, and technical approaches within each category. All prompts were run on 10 May 2026: ChatGPT through the OpenAI API without web search (so it returned no source URLs), and Perplexity and Google AI Overviews through automated browser sessions from a US endpoint. Every cited domain was recorded and classified as an official protocol source or one of seven third-party groups. An official source is a domain owned by one of the 50 sampled protocols.
Version 2 (17 October 2026) recounted the citation figures from the released dataset with that written definition. The headline citation totals changed (from 1,016 to 668 entries) and so did the official-page shares (from 1% and 4% to 8.8% and 7.7%). The central finding didn't change. The full change log is on the study page.
The full methodology, scoring formulas, and raw audit data are available in the downloadable dataset.
Independent corroboration
Two studies. One day apart. Same finding.
The 2026 AI Citation Visibility Study was published on 10 May 2026. On 11 May 2026, Ahrefs published an independent study on AI citation behaviour that reached the same primary conclusion: schema markup does not drive AI citations.
Neither study cited or referenced the other. The parallel finding emerged from separate methodologies applied to different datasets. This convergence is significant for any coverage of GEO, AI search, or structured data: the industry assumption that schema improves AI visibility now has two independent contradictions.
The 2026 AI Citation Visibility Study is the first peer-verifiable research to apply this analysis specifically to crypto protocols, where the data shows third-party sources doing most of the work of describing each protocol.
Chart pack
Charts for editorial use
All charts published under CC BY 4.0. Source attribution required: "CryptoContent.dev, 2026 AI Citation Visibility Study, DOI: 10.5281/zenodo.20146676"
Downloads
Study assets
Suggested citation
About the researcher
David Wood is a crypto content writer and researcher based in Scotland. His research focuses on how crypto protocols are represented in AI-generated answers and the structural factors that influence whether official sources are cited. The 2026 AI Citation Visibility Study is his first published empirical study in this area.
CryptoContent.dev is an independent content service working with DeFi and Web3 protocol teams on content designed for AI citation visibility.
Media contact
Get in touch
For interview requests, data queries, additional charts, or editorial questions about the study:
admin@cryptocontent.devResponses typically within one business day. Quotes, additional chart formats, and methodology clarifications available on request.
Quick facts for editors
Publication date: 10 May 2026
Current version: v2, 17 October 2026
Sample size: 50 protocols, 7 categories
Platforms tested: ChatGPT, Perplexity, Google AI Overviews
Citations analysed: 668 (Perplexity and Google AI Overviews)
Data access: Full dataset publicly available (CC BY 4.0)
DOI: 10.5281/zenodo.20146676
Independent corroboration: Ahrefs, 11 May 2026