How AI engines pick which crypto projects to cite.
An engine is not ranking pages, it is deciding who it can safely name. It builds that decision from sources it already trusts, then names the brands that recur across them. Which is why 92% of citations point at a domain the brand does not own, and why your own site tops out.
An engine is not ranking pages, it is deciding who it can safely name. It builds that decision from sources it already trusts, then names the brands that recur across them. Which is why 92% of citations point at a domain the brand does not own, and why your own site tops out near a tenth of your citations no matter how much you publish. Be present and consistent across the sources it reads, and you get named. See AI SEO for the wider picture.
The engine is asking: "who can I trust to name?"
An answer engine doesn't rank pages: it makes a judgment call. For a crypto query it's effectively asking: which projects are real, relevant, and safe enough to put my name behind? To answer that, it leans on signals that are hard to fake: who the independent press writes about, how consistently the web mentions you, whether your content is structured enough to quote, and whether you exist coherently across the places crypto buyers already trust.
That's good news. It means you can't buy your way in with a keyword, but you can earn your way in, and once earned, it's defensible. Below are the signals, in roughly the order they matter for crypto.
Five things every crypto answer is built on.
1 · Earned editorial coverage
Highest leverageIndependent crypto media writing about you is the strongest trust signal there is. Engines treat a CoinDesk or Decrypt mention as evidence you're real and noteworthy in a way your own blog never can be. Citation research repeatedly finds earned media drives several times more AI citations than owned content.
2 · Consistent brand mentions
Bigger than backlinksThe more places name you the same way, the more confident the model is that you're an established entity worth recommending. Unlinked mentions across the crypto web correlate more strongly with AI visibility than raw backlinks: the engine is gauging your footprint, not counting links.
3 · Structured, quotable content
Makes you liftableEngines cite passages they can lift cleanly: direct definitions, FAQs, comparison tables, schema-marked pages. If your key answers are buried in marketing prose, the model can't extract them, so it cites a competitor who made it easy.
4 · A coherent entity footprint
Crypto-specificReddit, YouTube, Wikipedia, CoinGecko, CoinMarketCap: when your project shows up consistently across the places crypto lives, engines read you as established. Gaps or inconsistencies (different names, missing listings) make the model hesitate to name you.
5 · Access for AI crawlers
The prerequisiteIf your site blocks GPTBot, ClaudeBot, or PerplexityBot (common on aggressively firewalled crypto sites) the engines literally can't read you. This is the easiest signal to fix and the one most often broken.
Want to know which of these five you're already winning, and which are costing you citations?
Run my free scan →The numbers point the same way.
Across independent studies of AI search behavior, and our own, the pattern is consistent: third-party trust beats self-promotion, and presence beats links. In our study of which crypto exchange AI engines recommend, YouTube was the single most-cited source and brand learn-hubs were the only owned pages engines cited at scale.
Where the citations actually landed
The five signals above describe how an engine chooses. This is what it chose, counted across every citation we captured.
| Source | Share of citations | What it costs to get on |
|---|---|---|
| The long tail (uncategorised domains) | 79.09% | Accumulated entity presence. Not purchasable |
| The brand's own domain | 7.7% | Free, and capped near this figure |
| Aggregators (CoinGecko, CMC, DefiLlama) | 2.26% | Listing application, free to low four figures |
| YouTube | 4.49% | Creator fees, or your own channel |
| Tier-2 crypto press | 2.27% | From roughly $800 a placement |
| Other community (Medium, Quora, Trustpilot) | 1.34% | Free, slow, unreliable to force |
| 2.19% | Free, hostile to anything that looks bought | |
| Tier-1 crypto press | 0.36% | From roughly $4,000 a placement |
| X / Twitter | 0.29% | Free, and almost never cited |
Two readings worth taking away. The engines pulled from 4,395 distinct domains, so there is no shortlist to buy your way onto. And the single largest identifiable source is the brand's own site at 7.7%, which is both the biggest lever you fully control and a hard ceiling on it.
Who occupies that long tail varies by vertical. In crypto payments it is largely the vendors themselves: five of the nine most-cited pages are published by payment companies ranking their own category. The numbers are in GEO for crypto payments.
You can see all of this for your brand.
citeOS turns these signals into a measured score. It probes 5 AI engines (ChatGPT, Perplexity, Gemini, Claude, and Google AI Mode) with 20 buyer-style prompts for your category, 100 observation points per scan, and reports where you're named, where you're cited as a source, which competitors take your place, and which of the five signals above are holding you back. Every weight is published in the open.
Then, on the PR side, citeOS scores 500+ crypto outlets on the exact qualities that earn AI citations, so when you invest in coverage, it lands where the engines actually look.
See the full scoring methodology →How AI citation works, answered.
What is the single biggest signal for AI citation in crypto?
Earned editorial coverage in credible crypto media. Citation research consistently finds that earned media drives several times more AI citations than a brand's own owned content, because engines treat independent third-party sources as more trustworthy than self-published claims.
Do backlinks still matter for AI citations?
They help, but plain brand mentions matter more for AI visibility. Studies show unlinked brand mentions across the web correlate more strongly with AI search visibility than backlinks alone. The model is building a picture of how established and relevant your entity is, not just counting links.
Can blocking AI crawlers hurt my crypto project?
Yes. If your site blocks GPTBot, ClaudeBot, or PerplexityBot (common on aggressively firewalled crypto sites) the engines can't read your pages and are far less likely to cite you. Allowing the major AI crawlers is a basic prerequisite for AEO.
How do I see which sources AI cites for my category?
Run a citation audit. Our 39,948-citation study shows what that looks like at scale. citeOS probes 5 AI engines with 20 buyer-style prompts and reports which sources each engine cites, where you appear, and where competitors take your place: 100 observation points per scan, entry scan free.
Related guides.
See which sources the AI cites instead of you.
Scan my brand free →