Updated Jul 8, 2026
You get cited in Google AI Overviews by being one of the three or more sources Google's Gemini model pulls into the synthesized answer, which is now a different game from ranking first. As of early 2026, AI Overviews appear on roughly 48 percent of US searches, and only about 38 percent of the URLs they cite still rank in Google's top 10, down from 76 percent a year earlier. That means most citations now come from pages outside the top 10, chosen for extractability rather than rank. The pages that get pulled lead each section with a direct, liftable answer, carry original data Google can attribute, and are grounded to a real entity through Wikidata and Organization schema. Source mix also matters: YouTube, Reddit, and Wikipedia dominate AI Overview citations, so genuine presence there feeds the overview. FORKOFF runs this as a $1,500 AEO sandbox audit that maps where you are already cited and the gaps to close.
The single biggest shift in AI Overviews is that rank stopped predicting citation. Ahrefs measured the share of AI-Overview-cited URLs that also rank in Google's top 10 falling from about 76 percent in mid 2025 to about 38 percent by early 2026, which means roughly six in ten citations now come from pages outside the top 10. AI Overviews appear on close to 48 percent of US searches, up from 6.49 percent a year earlier, so the surface is everywhere and it is choosing sources on extractability, not position. A page that ranks eleventh but leads with a clean, liftable answer can be cited over the page that ranks first but buries its point in marketing prose. The source table below shows where those citations actually come from.
An AI Overview is a synthesis of several sources, not a single winner. Across the data, 88 percent of AI Overviews cite three or more sources and only 1 percent cite a single source, so the realistic target is to be one of the three or four the model pulls, not the one page that beats everyone. That reframes the work from beating a competitor to being consistently present: the brand mentioned across the most cited sources is the one the overview repeats. It is also why source mix matters as much as your own page. YouTube, Reddit, and Wikipedia dominate AI Overview citations, so a brand with genuine presence on those surfaces feeds its own overview even before its owned pages get pulled.
When Google does pull a brand-specific claim, it needs to know which entity the claim belongs to. Without a Wikidata entity ID and Organization schema, AI Overviews paraphrase generically and conflate a brand with adjacent competitors, so the citation credit blurs. Grounding the brand to a real entity, a Wikidata ID with sameAs links to LinkedIn, Crunchbase, and the founder profile, plus Organization schema validated on every deploy, lets Google attribute the sentence to you specifically. Pair that with one original benchmark the overview can quote (methodology, numbered citations, a license that permits attribution) and you give Google a falsifiable claim with your name on it, which is exactly the kind of source a synthesized answer prefers. This is the entity-and-original-data core of FORKOFF's $1,500 AEO sandbox audit.
Where Google AI Overviews get their citations
| Source | Share of AI Overview citations | Why it matters |
|---|---|---|
| YouTube | About 23% | The single largest AI Overview citation source (Surfer SEO, 46M citations) |
| About 21% | First-hand community answers Google trusts for buying-intent queries (DemandSage) | |
| Wikipedia | About 18% | Entity grounding and definitional authority (Surfer SEO) |
| Brand and company sites | About 31% | Owned pages that lead with a liftable answer and original data (Presenc AI, 84K queries) |
Figures come from separate 2026 studies and are not additive. What holds across all of them: 88% of AI Overviews cite three or more sources and only 1% cite a single source, so being one of the committee, not the rank-1 page, is the target.
Rank position no longer predicts an AI Overview citation
| Metric | 2025 | 2026 |
|---|---|---|
| AI-Overview-cited URLs that rank in Google's top 10 | About 76% (Ahrefs, Jul 2025) | About 38% (Ahrefs, Mar 2026) |
| Citations from outside the top 10 | About 24% | About 62% |
| US searches showing an AI Overview | 6.49% (Jan 2025) | About 48% (Feb 2026) |
| Google searches ending without a click | Rising | About 68% (early 2026) |
Extractability and entity grounding now outweigh rank. Brands cited in AI Overviews earn about 35% more clicks even as overall click volume falls, so the citation is worth more than the position that used to earn it.
AI Overview optimization is structural, not a domain-rating game. The 12 on-page patterns that make a page extractable and citable, with first-party data.
Read
DeFi protocol marketing is a TVL problem. The 2026 playbook to take a protocol from zero to its first real Total Value Locked, backed by on-chain data.
Read
DePIN marketing is a distribution and trust problem, not a hardware one. The 2026 playbook to take a network from testnet to a token it keeps using.
Read