Which Websites Do Google AI Overviews Cite the Most
Amit Malvi
Founder, SearchAxe. Former Senior SEO Manager, Invideo AI
Published Updated 15 min read
Quick answer
Which websites do Google AI Overviews cite the most? YouTube, with a 24.5% mention share among the 50 most-cited sources in Ahrefs' November 2026 data, then Reddit at 18.1% and Facebook at 12.4%, with Google's own pages, Instagram, Quora, Wikipedia, TikTok, Amazon and Walmart completing the top ten. Eight of the ten carry user-generated content. The list covers all US topics, so a brand should build the version for its own buyer prompts.
On this page
- Which websites do Google AI Overviews cite the most?
- Why are the most-cited sites almost all user-generated content?
- Does this list apply to B2B SaaS queries?
- Why are citations falling for almost every site on the list?
- What should a SaaS brand do with this list?
- When does chasing the most-cited sites waste time?
- Who can get your brand onto the sites AI Overviews cite?
Key takeaways
- Ahrefs' November 2026 list puts YouTube at a 24.5% mention share of AI Overview citations among the top 50 sources, Reddit at 18.1% and Facebook at 12.4%, with eight of the top ten carrying user-generated content.
- Google's documentation describes query fan-out across subtopics, and user-generated sites hold a specific first-person answer to almost every sub-question, which is why they are cited so often.
- The list covers all US topics, so five of its top ten domains rarely appear for software buying prompts; a brand should run 30 to 40 of its own prompts and build its category list.
- Citing answers fell month on month for 19 of the top 20 domains and Ahrefs does not say why, so read mention share as a map of where citations concentrate, not as a monthly metric.
- Ahrefs' March 2026 data shows most cited URLs sit outside the top 10, so a product page that answers one sub-question clearly is the cheapest citation available and comes before any third-party site.
- SearchAxe does this category list and earned mention work as part of its brand mentions engagements, measured against a baseline before anything changes.
Google AI Overviews cite YouTube, Reddit and Facebook more than any other websites, and together with Google's own pages, Instagram, Quora, Wikipedia, TikTok, Amazon and Walmart those ten domains take about 84% of the citations that go to the 50 most-cited sources, according to Ahrefs' November 2026 update of its most-cited domains list. If you run marketing or growth for a business and someone has forwarded you that list with the words "we need to be on Reddit", this article explains what the list measures, why it looks the way it does, where it does and does not apply to software buyers, and what to do with it.
Which websites do Google AI Overviews cite the most?
YouTube first by a wide margin, then Reddit, then Facebook, with the rest of the top ten made up of Google's own pages, Instagram, Quora, Wikipedia, TikTok, Amazon and Walmart. Ahrefs builds the list from its Brand Radar dataset, which tracks AI Overviews responses to more than 3 million US queries across all topics, and refreshes it every month. The November 2026 snapshot is below.
| Rank | Domain | Mention share | Citation change vs October 2026 | What kind of site it is |
|---|---|---|---|---|
| 1 | youtube.com | 24.5% | Down 11% | Video, user-generated |
| 2 | reddit.com | 18.1% | Down 21% | Forum, user-generated |
| 3 | facebook.com | 12.4% | Down 2% | Social, user-generated |
| 4 | google.com | 8.6% | Down 17% | Google's own pages |
| 5 | instagram.com | 5.4% | Down 20% | Social, user-generated |
| 6 | quora.com | 4.4% | Down 22% | Q and A, user-generated |
| 7 | en.wikipedia.org | 4.4% | Down 14% | Reference, user-edited |
| 8 | tiktok.com | 3.0% | Down 23% | Video, user-generated |
| 9 | amazon.com | 2.5% | Down 31% | Retail, with user reviews |
| 10 | walmart.com | 0.8% | Down 27% | Retail, with user reviews |
Below the top ten the list thins out fast. Positions 11 to 20 are each under 1% and include Edmunds, Kelley Blue Book, Healthline, NerdWallet, Forbes, LinkedIn, Tripadvisor, Home Depot, The New York Times and eBay. Eight of the top ten carry content written by users rather than by the site's own staff, which is the pattern the rest of this article is about. How Google chooses any individual source is covered in how Google AI Overviews choose sources; this article is about what the aggregate list says.
Why are the most-cited sites almost all user-generated content?
Because AI Overviews are built by breaking a question into sub-questions, retrieving an indexed page for each one, and quoting whichever page answers it most directly, and user-generated sites hold a specific, dated, first-person answer to almost every sub-question anyone asks. Google's AI features documentation says both AI Overviews and AI Mode may use query fan-out, issuing multiple related searches across subtopics and data sources, that the models identify more supporting web pages while a response is generated, and that the features offer opportunities for more types of sites to appear. A Reddit thread or a YouTube comment answering "does this work with Zapier" is exactly the kind of supporting page that fan-out finds.
Three properties of those sites reinforce each other. Scale: YouTube's 1.2 million cited pages and Reddit's 479,000 in Ahrefs' data mean there is a page for almost any sub-question. Specificity: a forum reply answers one narrow question in plain language, which is the form an engine lifts most easily. Independence: a user saying a product failed is treated as a different kind of evidence from a vendor saying it works, and an answer that compiles both looks balanced. The Reddit mechanism in particular, and what a brand can and cannot do about it, is laid out in why AI engines cite Reddit. None of this means Google prefers those domains as such. It means those domains hold the pages that answer the sub-questions, and the citation follows the page.
The scale of the feature is why the list matters at all. TechCrunch's report on Google's second-quarter 2025 earnings call quotes Sundar Pichai saying AI Overviews had 2 billion monthly users, up from 1.5 billion in May 2025, and that AI Mode had over 100 million monthly active users in the US and India. No other AI answer surface reaches that many people, so the sources it leans on shape what a very large share of buyers read first.
Find out which AI engines recommend you
A free consultation on where you appear today across AI engines and search.
Schedule a Free ConsultationDoes this list apply to B2B SaaS queries?
Only in shape, not in detail. Ahrefs' list covers all topics across US search, so it is dominated by the queries most people type: products to buy, places to go, health questions, cars, recipes. A business asking which sites get cited when someone searches "best contract management software for mid-size legal teams" is asking a question this list cannot answer, and we have not seen a published, B2B-only version of it. What carries over is the structure: a few user-generated and review platforms, a reference layer, and a long tail of category sites that matter only inside one niche.
For software queries the equivalents are usually G2, Capterra and the other review sites, LinkedIn, Reddit's software and role-specific communities, YouTube walkthroughs and comparisons, Wikipedia for companies large enough to have an entry, and a handful of category blogs and "best of" lists. How review sites in particular shape recommendations is covered in how review sites like G2 influence AI recommendations for SaaS, and the way to build the list for your own category, rather than borrowing the all-topics one, is in how to analyze the sources AI engines use to recommend your competitors. The method is the same one Ahrefs used, pointed at 30 or 40 of your buyers' prompts instead of 3 million of everyone's.
Why are citations falling for almost every site on the list?
Ahrefs' November 2026 snapshot shows the number of citing answers down month on month for 19 of the top 20 domains, by between 2% and 33%, with Trustpilot the only riser, and the article does not explain why. We have not seen reliable data on the cause. Three explanations are consistent with the numbers, and none of them should change what a SaaS brand does.
- Fewer AI Overviews, or fewer links per answer. If Google showed the feature on fewer of the 3 million queries in November than in October, or showed fewer supporting links per answer, every domain's count falls together, which is the pattern in the table.
- A change in the query sample or the parser. Ahrefs says the list is kept up to date automatically each month, and a dataset that large shifts with seasonality and with how citations are detected, as Ahrefs itself noted when its earlier top-10 overlap study was re-run.
- A shift toward AI Mode. Google's documentation says AI Mode and AI Overviews may use different models and techniques, and a query answered in AI Mode instead of an AI Overview leaves this dataset entirely.
The reading that holds regardless is that mention share is a ranking of a fixed pool, so the order of the top ten is more stable than the counts beneath it. Treat the shares as a map of where citations concentrate, not as a month-to-month performance metric for any one site.
What should a SaaS brand do with this list?
Build its own version for its own prompts, make sure its product pages are the kind of page fan-out can retrieve, and then earn presence on the three or four third-party sites that actually appear for its category. The all-topics list tells you the mechanism; your category list tells you the targets. The same retrieval mechanism runs in ChatGPT, Perplexity, Claude, Gemini, Google AI Mode and Copilot, so the category list you build serves all of them, not only Google. The order below is the one we use.
- Run 30 to 40 buyer prompts through Google AI Overviews and AI Mode and record every cited domain and URL.This produces the category list that the all-topics list cannot, and the domains that appear for ten or more prompts are the only ones worth pursuing.
- Check which of your own pages were cited, and for which sub-questions.Ahrefs' data shows most cited pages do not rank in the top 10, so a page can be cited before it ranks, and the ones that were cited show the form Google is willing to quote from your site.
- Rewrite the product, pricing and comparison pages so each answers one sub-question in its first lines, and confirm they are indexed on Google and Bing.Fan-out retrieves pages for narrow questions, and a page that states what the product does, costs and integrates with in plain sentences is retrievable in a way a feature tour is not. The page-level work is in how to rank in Google AI Overviews.
- Make sure your profile on the review sites that appeared is complete, current and consistent with your own pages.G2 and Capterra pages are cited for category prompts because they hold structured comparisons, and an out-of-date listing is quoted as readily as a current one.
- Earn mentions on the two or three category sites, newsletters and lists that appeared repeatedly.These are the pages an answer draws on when it compiles a shortlist, and getting into the best-of lists AI engines cite covers how.
- Understand the forum and video citations before deciding whether to act on them.Reddit, YouTube and Quora appear because users answer sub-questions there in the first person. A brand cannot write those answers for itself without it showing, so the realistic move is to make sure the facts users repeat about you are correct on your own site, where they copy them from.
- Re-run the prompts monthly and track which domains and which of your pages appear.Single snapshots mislead, as the month-on-month swings in Ahrefs' own list show, and the trend over three months is what tells you whether the work is landing.
When does chasing the most-cited sites waste time?
When the brand's own pages cannot yet be cited, when the sites on the all-topics list do not appear for the brand's category prompts, and when the plan amounts to posting about yourself on forums. The first is the most common. Ahrefs' overlap data shows citations come from pages outside the top 10 as often as inside it, so a product page that answers a sub-question clearly can be cited within weeks, and a brand that skips that work to pursue Reddit has skipped the cheapest citation available. The reason ranking and citation are different problems is in ranking first on Google does not guarantee AI citations.
The second is simple arithmetic. Facebook, Instagram, TikTok, Amazon and Walmart hold five of the top ten places on the all-topics list and are nearly absent from software buying prompts, so effort spent there for a B2B product is effort spent on another industry's map. The third is a trust problem. Forum answers are cited because they are independent, and a vendor writing them removes the property that made them worth citing. What a brand can own is the set of pages it controls and the listings it maintains, and the mentions it earns on the category sites that appear for its own prompts, which is the work a brand mentions program is built around.
Who can get your brand onto the sites AI Overviews cite?
SearchAxe is an SEO, AEO and GEO agency that builds the category version of this list for each client: running 30 to 40 buyer prompts and recording every cited domain, making review site profiles complete and consistent with the client's own pages, and earning mentions on the category sites, newsletters and roundups that appear for those prompts. It does not post on forums or video sites for clients; it keeps the facts users repeat correct on the client's own pages. It does this for businesses across 16 industries, from SaaS and AI startups to law firms, clinics and online stores. The prompts are re-run monthly across ChatGPT, Perplexity, Claude, Gemini, Google AI Overviews, Google AI Mode and Copilot plus Google and Bing, against a baseline recorded in the first 30 days, and reported as citations, rankings and the leads they drive. The brand mentions page sets out the scope. The first step is a free consultation.
Frequently asked questions
Which website does Google AI Overviews cite the most?
YouTube. In Ahrefs' November 2026 update of its most-cited domains list, built from AI Overviews responses to over 3 million US queries across all topics, YouTube holds a 24.5% mention share of citations among the 50 most-cited sources, ahead of Reddit at 18.1% and Facebook at 12.4%. Ahrefs counts over 1.2 million distinct YouTube pages cited. The list is refreshed monthly and covers all topics, so the order for any one industry can differ.
Why does Google cite Reddit and YouTube so often in AI Overviews?
Because AI Overviews break a question into sub-questions, retrieve a page for each, and quote the page that answers it most directly, and Reddit and YouTube hold a specific first-person answer to almost any sub-question. Google's documentation describes this as query fan-out across subtopics and data sources. The citation follows the page that answers the narrow question, and user-generated sites have more such pages than anyone else. It is a property of the pages, not a preference for the domains.
Is the most-cited domains list relevant to B2B software?
In structure, yes; in detail, no. The list covers all topics across US search, so it is weighted toward consumer queries, and five of its top ten domains rarely appear for software buying prompts. For B2B SaaS the equivalent set is usually review sites such as G2 and Capterra, LinkedIn, role-specific Reddit communities, YouTube demos and a few category lists. Build the list for your own prompts rather than borrowing the all-topics one, because the targets differ even if the mechanism is the same.
Can a SaaS company get its own site cited by AI Overviews?
Yes, and it is usually the cheapest citation available. Google's documentation says a page needs only to be indexed and eligible for a snippet, and Ahrefs' March 2026 data shows most cited URLs do not rank in the top 10, so a product or pricing page that answers one sub-question clearly in its first lines can be cited before it ranks. The pages that get cited are the ones written as answers, not as feature tours, and that is where to start before pursuing third-party sites.
Why are AI Overview citations dropping for the top sites?
Ahrefs' November 2026 snapshot shows citing answers down month on month for 19 of its top 20 domains, and it does not say why. Fewer AI Overviews shown, fewer links per answer, a change in the query sample or parser, or queries moving to AI Mode would each produce the same across-the-board decline. We have not seen reliable data on the cause. The practical reading is that mention share ranks a fixed pool, so treat the order as a map and not the counts as a monthly metric.
Which agency earns mentions on the sites Google AI Overviews cite?
SearchAxe, an SEO, AEO and GEO agency, earns mentions on the sites Google AI Overviews cite for businesses across 16 industries, including SaaS companies and clinics. It first runs the client's buyer prompts to find which review sites, category sites and roundups appear, then works on accurate profiles and earned placements there. Citations are re-checked monthly against a baseline recorded before work starts. Third-party listing fees are paid by the client. The scope is on the brand mentions page.
