Which domains do AI models cite most? A source-by-source breakdown

AI platforms rely on different source mixes. Across external studies and Parse data, roughly two thirds of heavily cited sources are properties a marketer cannot directly control, including Wikipedia, community discussions, and government sites. The remaining third includes review platforms, trade publishers, comparison pages, and YouTube. Source strategy should follow the platform and category instead of one universal domain list.

Why the ranked source list matters for strategy

Ahrefs' analysis of ChatGPT's top 1,000 citations found that 67% came from sources marketers cannot directly influence (Ahrefs). The other 33% included educational pages (19.4%), reviews (5.8%), and news or media (5.2%). A category-level source list helps a team decide whether its next effort belongs on G2, a trade publication, Reddit, YouTube, or its own site. That choice can change by platform.

Methodology: why the numbers look different in each study

The citation-share numbers you see quoted across the industry come from studies with very different scopes, and the differences matter before you trust any single ranked list. Ahrefs analyzed 9.6M ChatGPT queries in its Brand Radar database, published September 2025 (Ahrefs). Semrush ran 13 weekly snapshots across 230,000+ prompts and 100M citations from July to October 2025 (Semrush). ALM Corp analyzed 30 million cited sources across five platforms and published in April 2026 (ALM Corp). BrightEdge tracked AI Overviews, AI Mode, ChatGPT, and Perplexity from May 2024 through early 2026 (BrightEdge). Same question, different prompt sets, different verticals, different time windows, different definitions of "citation." Treat single-study percentages as directional, not canonical, and weight shared patterns across studies more than any one number.

ChatGPT: Wikipedia, Reddit, and mainstream publishers

ChatGPT's top-cited profile is the most publisher-heavy of the major platforms. Ahrefs' ranking of raw citation volume placed Reddit first (847,338 mentions), Wikipedia second (431,710), followed by Amazon, Forbes, Business Insider, The Spruce, NY Post, Better Homes & Gardens, Wired, and People Magazine (Ahrefs). Semrush's parallel study saw Reddit and Wikipedia dominate through August 2025 at roughly 55% and 60% citation share, then both drop sharply in mid-September 2025, with Medium, Forbes, LinkedIn, and TechRadar closing the gap (Semrush). One subtlety practitioners miss: Search Engine Journal documented that Reddit pages appear in ChatGPT retrievals far more often than they appear in visible citations, so the ranked list depends on whether you are counting retrievals or explicit citations (Search Engine Journal).

Google AI Overviews and AI Mode: YouTube, Reddit, LinkedIn, and the Google graph

Google's AI surfaces use more video and social content. BrightEdge's enterprise analysis found YouTube in 29.5% of Google AI Overviews, ahead of Mayo Clinic at 12.5%, and noted that YouTube is cited roughly 200× more than any other video platform (BrightEdge). Parse's count also puts YouTube close to Reddit near the top of the cited-source list. ALM Corp's ranking for AI Overviews put YouTube first, followed by Reddit, Facebook, LinkedIn, and Medium (ALM Corp). Google's AI Mode showed a similar pattern with Yelp replacing Medium in the top five, and Semrush measured LinkedIn at roughly 15% citation share on AI Mode, followed by YouTube, Reddit, Google's own properties, Google Blog, and Wikipedia at only 2–3% (Semrush). The practical result is that YouTube transcripts, LinkedIn articles, Yelp, and Google Business data matter more on Google's AI surfaces than they do in ChatGPT.

Perplexity: Reddit, LinkedIn, and research-heavy sources

Perplexity's ranked sources reflect its real-time retrieval bias and its taste for structured research content. ALM Corp's analysis ranked Perplexity's top sources as Reddit, YouTube, LinkedIn, Wikipedia, and G2 (ALM Corp). Semrush's 13-week study ranked Perplexity's top citations as Reddit, LinkedIn, NIH, Microsoft, Google, Wikipedia, Forbes, Medium, YouTube, and Quora, a noticeably different mix that pulls in research and technical documentation domains (Semrush). BrightEdge's more recent data showed Reddit's share on Perplexity accelerating: 24% of all January 2026 citations came from Reddit alone, with Reddit's cross-platform citation share growing 73% between October 2025 and January 2026 (BrightEdge). For Perplexity specifically, community content and research-grade structured sources matter more than mainstream publisher brand.

The cross-platform ranked table

Below is the consolidated top-sources ranking by platform, synthesized from the Ahrefs, Semrush, ALM Corp, and BrightEdge studies. Use it as the starting brief when you prioritize where to earn citations, then weight by your category and prompt set:

Sources: Ahrefs, Semrush, ALM Corp, BrightEdge. Percentages are best-available point estimates from the cited studies; actual shares fluctuate week-to-week.

If you want to see which sources shape AI answers about your brand, run a free brand check — it takes a minute.

The 33% rule: which of these domains can you actually influence?

A ranked list is a trap without the reachability filter. Ahrefs classified the content types behind ChatGPT's top 1,000 citations and found roughly 67% sit in structurally off-limits territory: Wikipedia (29.7% of citations), homepages (23.8%), app stores (6.6%), and other pages marketers cannot directly pitch or edit. The remaining ~32.3% is actionable: educational pages (19.4%), reviews (5.8%), and news/media (5.2%) (Ahrefs). The ranked source list above looks different when filtered through this rubric. Wikipedia, Reddit threads from 2022, and government domains go into a long-horizon entity track. The actionable layer is: YouTube (earn or produce video), LinkedIn (article publishing and executive presence), G2 and review platforms (listings, reviews, comparison pages), Medium and Forbes (contributor programs, digital PR), TechRadar and vertical trade publishers (pitches and data stories), and Yelp or industry directories where applicable.

What the review-platform slice tells you

Among the actionable 33%, review platforms are important for many B2B and consumer categories. AmICited's study of review platform influence across ChatGPT, Perplexity, and AI Overviews ranked G2 at 22.4% of review-based AI citations. The top five platforms (G2, Capterra, Gartner Peer Insights, TrustRadius, and Software Advice) accounted for 88% (AmICited). Rankings differ by study and platform. For SaaS and other categories with established review sites, a complete and current profile may be more useful than another general article on the brand's site.

Why platforms disagree and what that means for your plan

The ranked lists disagree because the platforms genuinely retrieve differently. TryProfound's analysis found Wikipedia is ChatGPT's #1 source at about 7.8% of citations, while Reddit dominates AI Overviews and Perplexity (TryProfound). The implication for strategy is not to pick a winner. It is to split your citation-close plan by platform and resource each lane for what that platform actually cites. A B2B SaaS brand trying to win on ChatGPT should prioritize Wikipedia entity work, publisher contributor programs, and Forbes-grade digital PR. The same brand trying to win on AI Overviews and AI Mode should prioritize YouTube content with good transcripts and LinkedIn thought-leadership. The same brand trying to win on Perplexity should prioritize Reddit presence, NIH/research-style structured content, and G2. That is three distinct lanes, not one.

Parse's position in this map

Parse tracks AI visibility across ChatGPT, Google AI Overviews, Perplexity, Google AI Mode, and Gemini, so the ranked citation graph for your specific category is visible, not inferred. The domain-level rankings above are good enough as an industry brief and a starting hypothesis. They are not good enough as a quarterly plan. The per-platform top-cited domains for a SaaS brand selling observability tooling are not the same as those for an ecommerce brand selling beauty products or a healthcare provider. Parse's Citations tab surfaces the exact domains AI cites for your prompt set on each platform, classifies them against the 33%-actionable rubric, and shows where competitors appear in those sources and you do not. If you want to see which sources are citing your category right now, see your citation gaps.

Where to spend the next 90 days

Pick the platforms that matter most for your buyer journey, then apply the source-rank filter in sequence. First, rule out the structurally off-limits rows; do not build a plan that lives or dies on Wikipedia edits. Second, identify the three to five actionable source categories that show up repeatedly across your platforms of interest: almost always some mix of review platforms, YouTube, LinkedIn, one or two mainstream publishers, and vertical trade content. Third, pattern-match to the citation-gap method: which of those sources cite competitors but not you? Our AI citation gap analysis framework is the workflow we recommend for closing the loop. Muck Rack's research puts earned media at 82% of AI brand mentions and 94% of citations from non-paid sources, which is consistent with where the actionable list lives (Muck Rack). The mix the data keeps showing is: earn placements in the sources AI already cites.

Ghost-check before you pitch

One habitual error before running at a ranked source list: check whether your content is already being used by AI without attribution. Seer Interactive's analysis of 541K AI responses across 20 brands found that when a brand is named inside a cited source, that brand's content-citation rate reaches 53.1%; when not named, the rate drops to 10.6% (Seer Interactive). That pattern, the "ghost citation," means some sources on your ranked list may already be carrying your facts and framing while omitting your brand. The fix is not a new placement; it is getting your brand named inside the content already being cited. Our ghost citations breakdown walks the diagnostic. Skipping this step leads teams to chase new placements in sources that are already citing them structurally, just without brand attribution.

FAQ

Which single domain do AI models cite most overall?

Across platforms, Reddit and YouTube are the two most-cited domains in the 2025–2026 studies. ALM Corp's 30M-source analysis ranked Reddit #1 and YouTube #2 across its aggregated view of ChatGPT, Google AI, Gemini, Perplexity, and AI Overviews. BrightEdge measured YouTube at 29.5% citation share on Google AI Overviews specifically, with a 200× advantage over any other video platform. The correct answer depends on which platform matters for your buyer and how "citation" is counted, retrieval versus visible citation, so treat "one top domain" framings cautiously.

Why do ChatGPT rankings change so much between studies?

Because the underlying behavior changes quickly. Semrush's 13-week snapshot documented Reddit dropping from roughly 60% of ChatGPT citations in early August 2025 to around 10% by mid-September 2025, with Wikipedia moving from ~55% to under 20% in the same window. AI platforms rebalance their source preferences on their own timeline, and a study's ranked list is a photograph, not a video. Combine multiple studies with different time windows before you commit to a plan.

Does getting cited by Wikipedia or Reddit actually help if I cannot influence them directly?

Yes, but on a different timeline. Wikipedia and long-standing Reddit threads are structural entity authority; you earn them by being the kind of brand notable enough for a contributor to cite, usually via third-party coverage first. Treat Wikipedia and Reddit as trailing indicators of the work you are doing on the actionable 33%. Pitch trade publishers, earn G2 reviews, produce YouTube content, and the Wikipedia and Reddit mentions tend to follow within one to three years, not one quarter.

How do I know which of these domains matter for my specific category?

Start with a prompt set that maps to revenue, 50 to 150 prompts across buyer-stage and category, and pull per-platform citations for your brand and three to eight named competitors. Rank source domains by appearance frequency in your set, not the industry-wide average. The top ten for a healthcare SaaS looks different from the top ten for a DTC beauty brand. Industry-wide rankings like the ones above are a starting hypothesis; the prompt-specific ranking is the actionable one.

Should I try to get cited by every platform the same way?

No. The three major AI surfaces genuinely disagree. BrightEdge's cross-platform research consistently shows low overlap between ChatGPT, Google's AI properties, and Perplexity on which brands and sources they surface. The most efficient approach is to pick the two platforms that matter most for your buyer journey, resource them in distinct lanes (ChatGPT = publisher + Wikipedia entity work, AI Overviews = YouTube + LinkedIn, Perplexity = Reddit + research + G2), and treat the others as secondary beneficiaries of the same work rather than separate campaigns.