Which sources do ChatGPT and Perplexity cite most for B2B questions?

For B2B questions, the sources ChatGPT and Perplexity cite concentrate mostly on a small, recurring group: encyclopedic reference sites like Wikipedia, community platforms like Reddit, independent review and comparison sites, specialised trade media and official supplier documentation. Your own website is added to the mix, but rarely as the only voice. In this article you will read which domain types keep surfacing, why AI trusts them, how ChatGPT and Perplexity differ, and therefore where you need to be present yourself. This is a core part of generative engine optimization.

Which domain types keep coming back?

AI citations cluster around a handful of recognisable domain types, not around a long random list. When you put several citation analyses side by side, you see the same categories float to the top every time, regardless of the exact topic.

The recurring types are:

• Encyclopedic and reference sources. Wikipedia and comparable reference sites appear strikingly often, because they summarise facts in a structured and neutral way. AI likes to use them as an anchor point for definitions and background.
• Community and discussion platforms. Reddit and forums carry a lot of weight, especially for questions about experiences, pros and cons or “what actually works”. AI reads those threads as authentic, human-validated input.
• Independent review and comparison sites. For software choices and supplier comparisons, AI leans on platforms that place products side by side. Think of G2 and Capterra in the software landscape.
• Trade media and industry publications. News and trade titles that follow a market or field provide context and figures that AI treats as authoritative.
• Official and primary sources. Documentation, product pages and research from the source itself are used for hard specifications and facts.

An important nuance: the exact share per domain varies strongly by study, by question and by moment. The stable finding is not a precise percentage, but the pattern. A limited set of source types captures the lion’s share of citations. We call that citation concentration, and it is precisely why publishing content at random yields little.

Why does AI trust exactly those sources?

AI cites sources it can recognise as reliable, findable and unambiguous, not the source that shouts loudest. Three mechanisms explain the choice.

First, authority through consensus. If several independent places say the same thing about you or your category, the model treats that as a reliable signal. That is why external mentions often carry more weight than a nicely written sentence on your own site. We work this out further in why brand mentions carry more weight than backlinks.

Second, extractability. Sources that answer a question directly, factually and in a clear structure are easier to cite than slabs of marketing text. Encyclopedic and documentation sites naturally score well there.

Third, verifiability. Especially when a model searches the web live, it picks pages it can retrieve and check at that moment. How AI determines what is true and what it dares to cite, you can read in grounding: how AI determines what is true.

The consequence for B2B is concrete. AI likes to look for outside confirmation before it names a supplier in an answer. Your own website is needed as a foundation, but rarely enough as the only source.

Do ChatGPT and Perplexity differ in their source choice?

Yes, and the difference lies mainly in how fresh the sources are. It pays to nuance your approach slightly per platform.

Perplexity is built around live web search. For almost every question it retrieves a set of current pages and places visible, numbered sources underneath. As a result, recent, easily findable and crawlable pages carry a lot of weight here, and it is striking how strongly community content counts. How to get into that list of citations, we cover in getting found in Perplexity.

ChatGPT mixes two things: knowledge from the trained model and, when it switches on web search, live-retrieved sources. For factual or current questions it leans on reference and authoritative sources it can find again. For more general questions it draws more heavily on what the model already “knows”, where established, often-cited brands have an advantage. For the practical side, see getting found in ChatGPT.

The common thread stays the same across both platforms: user discussion and reference works rank strikingly high, often above the classic SEO winners. Reddit is the clearest example of that, and why that is and what you do with it, you can read in how Reddit makes your brand visible in ChatGPT.

What does this specifically mean for B2B questions?

For B2B questions, the weight shifts toward independent comparison and domain expertise, because the stakes and risk of a choice are higher. Someone looking for a supplier or software package wants proof, not a brochure.

In practice you see three source types recur often for B2B:

• Review and comparison platforms for “best tool for X” or “alternatives to Y”. Here products are placed side by side, exactly what a buyer, and therefore the AI, wants.
• Trade media and industry publications for market trends, definitions and context around a field.
• Community threads where professionals share what works in practice and what falls short.

That has a sober consequence. You can perfectly optimise your own website and still not be mentioned, simply because the outside confirmation is missing. For B2B your work is therefore twofold: make your own pages citable, and make sure the external sources AI trusts mention you correctly and consistently. How to align this with a business buying journey, we work out in GEO for B2B.

An honest caveat belongs here. You have no control over what a review site or a forum member writes, and you cannot buy or guarantee citations. What you do steer is your presence, your consistency and the quality of the facts you send out into the world.

How do you use this without chasing loose citations?

Steer toward presence in the source types that keep coming back for your topic, not the individual mentions you cannot control anyway. That is the shift from vanity goals to something that can genuinely influence leads and revenue.

In practice that means:

• Map your own source landscape. Ask ChatGPT and Perplexity the questions your buyers ask and note which domain types and concrete sites keep surfacing. That tells you where you need to be.
• Make your own pages extractable. Answer one clear question per page, at the top, in plain language. Make sure a crawler can read you smoothly.
• Build external confirmation. Be present and correctly mentioned on the relevant review, comparison and trade-media sources, and make sure your name, claims and figures are consistent everywhere.
• Measure qualitatively, not obsessively. Track whether you appear in the right source types and whether you are mentioned correctly more often, instead of counting on one chance citation.

If you would rather not figure this out yourself, we map which sources AI cites for your market and where the gaps are through our AI visibility.

The short summary

For B2B questions, ChatGPT and Perplexity cite a small, recurring circle: reference sites, community platforms, independent review and comparison sites, trade media and official documentation. The exact percentages differ by study, the pattern does not. AI looks for outside confirmation, so your own website is needed but rarely enough. Perplexity leans more heavily on fresh web results, ChatGPT mixes memory and live sources, and on both, user discussion ranks high. The win is not in chasing loose citations, but in structural presence in the source types that truly matter for your topic.

Book your free intake and we will look together at which sources AI cites for your market today.