Blog
AI Overviews for Crypto Queries: How Google Picks Citations
Inside Google AI Overview's citation logic for cryptocurrency queries — which sources get cited, why crypto media dominates, and what a brand can do to enter the citation graph.
Google AI Overview (AIO) shows up on roughly 38% of crypto-related queries we panel monthly — higher than any other vertical we measure. For “what is”, “how does”, “is it safe”, and “best of” queries, AIO sits at position zero and answers the question without a click. If your brand is not in the citation strip below the answer, the click is not coming.
This post is what we know about AIO’s citation logic for crypto specifically, based on a 12-month observation of 240 queries against a panel of 47 crypto domains.
Quick facts
| Parameter | Value |
|---|---|
| AIO presence on crypto queries | ~38% (monthly average, our panel) |
| Citations per AIO answer | 3–5 sources, occasionally 7 |
| Source-graph dominance | Cointelegraph, Decrypt, The Block, CoinDesk together = ~52% of all citations |
| Average citation refresh | ~quarterly; older citations rotate out |
| Time-to-first-citation for new brand | 8–16 weeks after first published article |
| Citation persistence | Once cited, brand stays cited for similar queries 60–80% of the time |
Which crypto sources does AIO cite most?
Top tier (cited on >5% of crypto AIO answers in our panel): Cointelegraph, CoinDesk, Decrypt, The Block, Bitcoin Magazine, Investopedia (for definitional queries), Forbes/Forbes Advisor, Reuters, Bloomberg. These are the publications a generic-news-aware LLM would rank highly anyway, and Google’s AIO inherits that bias through its training data and live retrieval.
Mid tier (1–5%): Crypto Briefing, Blockworks, BeInCrypto, U.Today, Cryptoslate, plus jurisdictional regulators (FINMA, MAS, BaFin, FCA), and key exchanges’ own knowledge bases (Binance Academy, Coinbase Learn, Kraken Learn).
Long tail (single-digit appearances): brand-owned content, well-cited Medium posts, GitHub READMEs for protocols, academic papers (arXiv) for technical queries.
Why does AIO cite the same publications repeatedly?
Three factors compounding. Authority recency — AIO’s retriever leans on freshness for crypto because the niche moves fast. Publications that publish 5–20 articles per day stay fresh; static sites don’t. Even a great pillar page gets out-cited by a worse Cointelegraph article published last week.
Citation graph density — AIO appears to weight sources that other already-cited sources link to. Cointelegraph links to The Block; The Block links to CoinDesk; AIO sees a tight cluster of mutual citations and trusts them as a group. Outsiders need to break in via at least one inbound link from this cluster.
Editorial structure — these publications follow predictable structures (lede, factbox, body) that match what AIO’s extractor wants. A page that buries the answer in the seventh paragraph loses to one that states it in the first 30 words. The structural pattern is what GS Playbook v4.3 codifies.
How does a brand actually get cited?
The realistic path for a brand site (not a publication) takes 12–24 weeks and three layers. Layer 1 — your own pages get cited as a secondary source: 8–16 weeks after publishing structured pillar content with FAQ schema, AIO will start citing you for long-tail queries inside your jurisdiction or category.
Layer 2 — you get named in 2–3 tier-1 crypto publications: 12–18 weeks. This requires PR pitching, original research (data studies move fastest), or expert quotes via HARO/Featured/Qwoted-style services. Once Cointelegraph or Decrypt names your brand, AIO starts associating you with the topic.
Layer 3 — your team members become cited individually: 18+ weeks. AIO sometimes cites individual experts by name when the article carries a real author with a verifiable LinkedIn and sameAs schema. For crypto specifically, having a named CSO or CCO who publishes regularly creates an expert-level citation surface that the brand site alone can’t reach.
What kills AIO citation chances?
Five quick disqualifiers we audit for. No real authors — pages with “Admin” or “Editorial Team” bylines are systematically demoted. Word salad disclaimers — “this is not financial advice” wrapped around every paragraph dilutes the answer; AIO can’t extract from a hedged page. Schema mismatches — FAQPage schema where the visible Q&A doesn’t match the JSON-LD; AIO ignores the whole block. Pseudonymous or anonymous team — the brand exists outside the trust graph. Domain age + content velocity mismatch — a 6-month-old domain with 3 articles will not break into citations regardless of quality; sustained publishing is a prerequisite signal.
Frequently asked questions
Does AIO citation drive measurable traffic? Yes but less than ranking position 1. Citation strip CTR is 6–9% of the AIO impressions; a position-1 organic listing is 30–35%. AIO citation is more about attribution than direct traffic — being named in the answer builds brand recall.
Can we pay to be cited? No. Sponsored content is not eligible for AIO citation regardless of how it’s labelled. The only paid path is paying for legitimate PR placement in publications already in the citation graph.
How do we measure AIO citations at scale? Monthly prompt panel — 30–60 prompts derived from your keyword cluster, run against Google AIO, with the answer + sources logged. We use a Supabase + Looker Studio setup for clients on the Authority retainer; for smaller scope, manual logging in a spreadsheet works for the first 6 months.
Does Schema.org Article author with sameAs actually help?
Yes. We A/B-tested across 14 jurisdiction pages on cryptolicense.pro. Pages with named author + LinkedIn sameAs got cited 1.8× more than pages with no author over a 90-day window.