QBiz Leads AI

Each AI Engine Has a Favourite Type of Source: How to Become One

Summary

Each AI engine reads the web and builds its answer from a small set of sources it trusts, and the three big engines trust different ones. ChatGPT reaches for authoritative, encyclopaedic pages (Wikipedia, sector bodies, established media). Google AI Overviews lean on community and video (Reddit, YouTube, your Google Business Profile). Perplexity is built heavily on Reddit and reviews. To get into the answer, be present and accurate wherever your engines read: get the authoritative listings right for ChatGPT, gather steady reviews for Perplexity, and add genuine community and video for Google, with your core facts identical everywhere.

In brief

  • Every engine writes its answer from a handful of trusted sources, not from your homepage.
  • ChatGPT favours authority: Wikipedia, sector registers and directories, established local and trade press.
  • Google AI Overviews favour community and video: Reddit, YouTube, and your Google Business Profile for local questions.
  • Perplexity leans hardest on Reddit and reviews, and behaves most like classic keyword search.
  • One broad, accurate, well-reviewed presence feeds all three: you rarely need a separate strategy per engine.

When someone asks ChatGPT, Google's AI answers or Perplexity to recommend a business like yours, the tool does not pluck a name out of thin air. It reads the web first, then writes its answer from a handful of sources it trusts. Get into those sources and you get into the answer. Stay out of them and you are invisible, no matter how good your website is.

The three big engines do not trust the same sources. They have genuinely different habits. ChatGPT reaches for encyclopaedic, authoritative pages. Google's AI Overviews lean on community and video. Perplexity is built almost entirely on one community site. If you treat them as one audience, you waste effort. If you understand what each one favours, you can aim your work where it actually counts.

This is a companion to our complete guide to Answer Engine Optimisation for local businesses: start there for the full picture, come here for how the engines differ.

Why "where the AI reads" matters more than "what your website says"

Most owners assume AI visibility is a website job: write the right words on your own pages and the AI will find you. Your website matters, but it is only one input. The engine is reading the whole web about you, and it weights some places far above others.

This shift is already mainstream, not a future worry. AI-generated summaries now sit on top of a large share of everyday searches, and when one appears people click through to a normal result far less often. Ahrefs measured a 34.5% lower click-through rate for the top-ranking page when an AI Overview is present, across a sample of 300,000 keywords (Ahrefs, AI Overviews Reduce Clicks by 34.5%)[1]. People are reading the answer and stopping. So the question is no longer "where do I rank?" It is "which sources does the answer get built from, and am I in them?" The sources differ by engine, so let us take them one at a time.

ChatGPT: it trusts authority and the encyclopaedia

ChatGPT behaves like a careful researcher who reaches for a reference book first. Its single most-cited source by a wide margin is Wikipedia.

A Semrush study that tracked more than 100 million AI citations in weekly snapshots between July and October 2025 found that, even after a sharp drop in how often ChatGPT leaned on them, "Reddit and Wikipedia are still ChatGPT's two most-cited domains" (Semrush, The Most-Cited Domains in AI, 2025)[2]. An independent analysis of 30 million cited sources by Peec AI, reported by Search Engine Land, points the same way: ChatGPT "favored Wikipedia, Reddit, and editorial sites like Forbes"[3].

After Wikipedia, ChatGPT's top sources are a mix of established media and review-style sites. The thread running through them is perceived authority and editorial standing, not raw popularity. That preference shows up in the numbers: a SE Ranking study of 129,000 domains and more than 216,000 pages found that sites with a Domain Trust score above 90 earned close to four times as many ChatGPT citations as sites below 43 (roughly 8.4 citations per page at the very top versus about 1.6 at the bottom)[5].

The Semrush data also shows how quickly this can move: ChatGPT reweighted its top sources in mid-September 2025, which is a reminder that these are habits, not fixed rules.

What this means for a local business

You will not out-Wikipedia Wikipedia, and as a single-location business you almost certainly will not get a Wikipedia page at all. That is fine, and chasing one is a waste of time. The play is different:

Google AI Overviews: it trusts the crowd and the video

Google's AI Overviews behave almost in reverse. Where ChatGPT reaches for the reference book, Google reaches for the conversation and the demonstration.

The same 30-million-source analysis found Reddit to be the most-cited domain across the major engines, with YouTube and LinkedIn also in the top five, and noted that Google in particular "leaned toward platforms like Facebook and Yelp"[3]. The Semrush data adds that Google's AI Mode has "a clear preference for Google's owned domains" such as YouTube and Google's own properties[2]. Wikipedia, which is a top source for ChatGPT, barely features in Google's picture. Google is pulling from a balanced mix of community discussion, video and professional networks rather than deferring to one encyclopaedic source.

That makes intuitive sense. Google owns YouTube and has a longstanding deal to surface Reddit content, and its whole search history is built on weighing many signals rather than deferring to one. The takeaway for you is that real, human, demonstrable content travels well in Google's AI answers.

What this means for a local business

Perplexity: it is built on Reddit

Perplexity is the most extreme of the three. It is, to a striking degree, a Reddit reader with a polished interface.

The independent 30-million-source analysis singled Perplexity out as emphasising "Reddit, LinkedIn, and G2 for B2B queries", with review platforms such as Yelp and G2 appearing often in recommendation queries[3]. Semrush's tracking similarly puts Reddit among Perplexity's top sources throughout its study[2]. The picture is clear: Perplexity wants community discussion and peer recommendation above almost everything else.

There is a useful flip side. Of the three engines, Perplexity behaves the most like classic search, matching the words in your content closely. That means the traditional SEO work you may already have done is not wasted on Perplexity; clear, keyword-clear pages still help you there. We cover that in our Perplexity playbook.

What this means for a local business

The pattern underneath all three

Step back and a single lesson emerges. Every engine, despite its different favourites, is doing the same thing: trusting a small set of sources it considers reliable for the question, then building an answer from them. None of them is mainly reading your homepage. (This is also why a household name does not automatically win the answer when a smaller, better-sourced business is the more relevant match.)

That is why AEO is not a website-only job. What actually counts is spread across the places the engines actually read:

A local business that is accurate on the authoritative sources, genuinely present in the relevant communities, steadily reviewed, and clear on its own pages is feeding all three engines at once. You do not need a separate strategy per platform so much as a presence broad enough that whichever source an engine reaches for, you are in it.

A simple cross-engine plan

You do not have to do everything. Ordered roughly by impact for a typical local business:

  1. Claim and complete your Google Business Profile. This single asset feeds local answers across Google directly and informs the others indirectly. It is step zero.
  2. Get accurate on the authoritative sources for your sector. List on the recognised industry body and the directories that matter in your trade, with details that match your website exactly. This is your ChatGPT play.
  3. Build steady, recent reviews on the platforms that count in your trade. This is your Perplexity play and a strong signal everywhere.
  4. Add a little genuine video and genuine community participation. A few plain-spoken videos and helpful forum answers cover the Google and Perplexity community appetite.
  5. Make your own pages answer real questions in plain words. Service pages and area pages phrased the way customers ask, which helps Perplexity match you and gives every engine something clean to quote.
  6. Keep your core facts identical across the whole web. Name, address, phone and services the same everywhere, so no engine has a reason to doubt who you are.

Common mistakes

Frequently asked questions

Why does ChatGPT cite Wikipedia so much but Google barely does?

They have different design philosophies. ChatGPT reaches for authoritative, encyclopaedic content as a baseline, so Wikipedia remains one of its two most-cited domains (Semrush, 2025)[2]. Google's AI Overviews lean on community discussion, video and professional networks, where Reddit, YouTube and platforms like Facebook and Yelp lead, and Wikipedia barely features (Search Engine Land, reporting Peec AI)[3].

Do I need a different strategy for each engine?

Not really. The smarter approach is a presence broad enough that whichever source an engine prefers, you are in it: accurate on authoritative sources for ChatGPT, present in communities and video for Google, steadily reviewed for Perplexity. One broad foundation feeds all three.

Does the AI keep my town when it searches?

Generally yes. OpenAI's own help documentation explains that ChatGPT search rewrites your question into targeted queries and uses your general location to improve results, giving the example of turning "good restaurants near me" into a city-specific search (OpenAI Help Center, ChatGPT Search)[4]. Location is the signal these tools hold onto, which works in a local business's favour.

Is Reddit really that important?

For Google AI Overviews and especially Perplexity, yes. An independent analysis of 30 million cited sources found Reddit to be the single most-cited domain across the major AI engines, with Perplexity leaning on Reddit particularly hard (Search Engine Land, reporting Peec AI)[3]. The key is to take part genuinely and openly, never to fake it.

My business is tiny. Can I really show up next to bigger names?

For local questions, yes. The engines weight location and reviews heavily, and you only have to be the clearest, best-documented answer in your own town, not the whole country. That is a far smaller hill to climb than competing nationally.

What actually makes ChatGPT more likely to cite a source?

Perceived authority, not popularity. A SE Ranking study of 129,000 domains found that sites with a high Domain Trust score were cited far more often, close to four times as much above a score of 90 as below 43. That is why the play for a local business is to be accurate and consistent on the authoritative sources that already carry weight in your sector, rather than relying on your own site alone.

Which single asset helps most across all three engines?

For local questions, a complete and current Google Business Profile. It feeds Google's local answers directly and gives every engine a consistent set of core facts (name, address, phone, services) to confirm you against. It is the step-zero asset before anything else.

Does video really matter, or is that just for Google?

It leans towards Google, where YouTube is a favoured source, but a plain-spoken sixty-second video also gives Perplexity something genuine to cite. It does not need production polish: it needs to be real, clearly labelled with who you are and where you operate, and answer a question a customer actually asks.

If my details differ slightly across the web, does it really matter?

Yes. Every engine is assembling a confident picture of who you are, and an old phone number on one directory against a new one on another quietly lowers that confidence. A less sure engine cites you less often, so keeping your name, address, phone and services identical everywhere is low-effort, high-value work.

Is my own website pointless for AI visibility, then?

No, it is just one input rather than the whole job. Clear service and area pages phrased the way customers ask give every engine something clean to quote, and they help Perplexity match you directly because it reads your words closely. The point is that your website works alongside the off-site sources each engine trusts, not instead of them.

Where to start

If you do only one thing this week, complete your Google Business Profile. After that, pick the engine your customers use most and feed its favourite source: authoritative listings for ChatGPT, reviews for Perplexity, a little video and genuine community presence for Google.

If you would rather see exactly which engines name you, which name your competitors, and which sources they are drawing on, that is what a QBiz AI Visibility audit does. We map which source each engine leans on for your category, show where you are absent from those sources, and rank the gaps worth closing.

Get your AI Visibility audit →

Sources

  • [1] Ahrefs, "AI Overviews Reduce Clicks by 34.5%": https://ahrefs.com/blog/ai-overviews-reduce-clicks/ (independent; analysis of 300,000 keywords; the presence of an AI Overview correlated with a 34.5% lower average click-through rate for the top-ranking page vs comparable informational keywords without one)
  • [2] Semrush, "The Most-Cited Domains in AI: A 3-Month Study," 2025: https://www.semrush.com/blog/most-cited-domains-ai/ (independent; 230,000+ prompts across ChatGPT, Google AI Mode and Perplexity in weekly snapshots 14 July–12 October 2025, over 100 million AI citations; "Reddit and Wikipedia are still ChatGPT's two most-cited domains"; Google's AI Mode shows "a clear preference for Google's owned domains"; Reddit is among Perplexity's top sources)
  • [3] Search Engine Land, "AI search engines cite Reddit, YouTube, and LinkedIn most: Study," reporting Peec AI's analysis of 30 million cited sources: https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138 (independent; Reddit is the most-cited domain across ChatGPT, Google AI Mode, Gemini, Perplexity and AI Overviews; ChatGPT "favored Wikipedia, Reddit, and editorial sites like Forbes"; Google "leaned toward platforms like Facebook and Yelp"; Perplexity "emphasized Reddit, LinkedIn, and G2 for B2B queries")
  • [4] OpenAI Help Center, "ChatGPT Search": https://help.openai.com/en/articles/9237897-chatgpt-search (primary; ChatGPT rewrites a question into targeted queries and uses general location from IP address to improve results, e.g. "good restaurants near me" becomes a city-specific search)
  • [5] Search Engine Journal, "New Data: The Top Factors Influencing ChatGPT Citations," reporting SE Ranking's study of 129,000 domains and 216,524 pages: https://www.searchenginejournal.com/new-data-top-factors-influencing-chatgpt-citations/561954/ (independent; domains with a Domain Trust score above 90 earned nearly four times the ChatGPT citations of sites below 43, roughly 8.4 citations per page at DT 97–100 versus about 1.6 below DT 43)

Leave a comment

Thoughts on this post? Leave a comment below. Comments are moderated before they appear, so yours will not show on the page straight away.

Your email is used only to contact you about your comment if needed — it is never published.

Comments

No comments yet. Be the first to leave one above.