Product Updates

Sources, Citations and Influence: What Shapes ChatGPT’s Answers About Your Brand

Geoffroy Guilhaume

•

•

•

7 min de lecture

TL;DR

  • ChatGPT and other AI engines build answers from two layers: their training data and the pages they consult in real time.

  • Each AI has favorite sources: Wikipedia dominates in ChatGPT, Reddit in Perplexity, Reddit and YouTube in Google AI Overviews.

  • Brand mentions across the web weigh roughly three times more than backlinks for AI visibility.

  • Identifying the pages that feed answers in your category is the first step to influencing what AI says about you.

Which sources influence ChatGPT’s answers? The direct answer

ChatGPT’s answers are influenced by two types of sources: the texts seen during model training (what it “knows”) and the web pages it consults in real time when it searches (what it “cites”). Among those pages, ChatGPT favors knowledge bases and established media, with Wikipedia at the top.

Model memory vs. real-time search

Model memory reflects your brand’s historical presence on the web. The more consistently your brand has been mentioned in recognized sources, the more the model spontaneously associates it with your category.

Real-time search depends on the pages the underlying search engine surfaces when the question is asked. Both layers matter. A brand missing from model memory can catch up through good, fresh sources. Conversely, a well-known brand can be described wrongly if the pages consulted are outdated.

What brands actually control

The share of sources a brand controls depends on the type of question.

For local and practical queries, Yext examined 6.8 million citations across ChatGPT, Gemini and Perplexity and found 86% originate from brand-controlled sources, including websites, business listings and review platforms. techstrong

For broader product research, McKinsey sees the opposite: a brand’s own website supplies only 5–10% of the sources AI engines consult when building answers.

Key takeaway For a practical question (opening hours, pricing, address), your website and listings do most of the work. For a choice question (“best X for Y”), media, reviews and communities decide. A source strategy must cover both.

Which types of sites do AI engines favor? Source comparison

Each AI has its own source “diet.” The reference study comes from Profound, which analyzed tens of millions of citations between August 2024 and June 2025.

Platform

#1 source

Other strong sources

Profile

ChatGPT

Wikipedia (47.9% of top 10)

Reddit (11.3%), Forbes (6.8%)

Encyclopedic and established media

Google AI Overviews

Reddit (21%)

YouTube (18.8%), Quora (14.3%), LinkedIn (13%)

Communities and video

Perplexity

Reddit (46.7% of top 10)

Varied sources, press

Community-driven and recent

Source: Profound, share of citations among each platform’s 10 most-cited domains.

According to Profound, within ChatGPT’s top 10 most-cited sources, Wikipedia accounts for nearly half (47.9%) of citations. tryprofound

Google’s AI Overviews lean more on community and video content: Reddit at 21%, YouTube at 18.8%, Quora at 14.3% and LinkedIn at 13%. clickguard

Overlap between platforms is low: research firm 5W estimates that only 11% of domains are cited by both ChatGPT and Perplexity. 5wpr

Preferences change fast

Citation habits aren’t fixed. ChatGPT’s Reddit citation share collapsed from roughly 60% to 10% in mid-September 2025 before stabilizing. A strategy built on a single source or a single platform is fragile by nature. 5wpr

Key takeaway There is no single source strategy for AI, only one strategy per engine. ChatGPT values encyclopedic and editorial authority, while Perplexity and Google AI Overviews value community proof. Measure each engine separately.

Reddit, YouTube, Wikipedia, media and forums: their influence on ChatGPT and brand visibility

Third-party platforms play a double role. They get cited directly in answers, and they shape how the model perceives your brand.

Reddit: the social proof that weighs most

Reddit is the community source AI engines cite most. A thread where users compare your product with a competitor’s can be reused in an answer, tone included, positive or negative. A single heavily cited critical thread can weigh on recommendations for a long time.

YouTube: an increasingly strong signal

YouTube is the second-largest source for Google AI Overviews. AI engines transcribe and use product reviews, tutorials and video comparisons. A December 2025 Ahrefs analysis found YouTube mentions among the signals most correlated with brand visibility in AI.

Wikipedia: ChatGPT’s trust anchor

Wikipedia serves as a factual reference: founding date, business activity, leadership, products. A missing, incomplete or outdated entry directly affects what ChatGPT states about your company. Wikidata, its structured counterpart, plays the same role for entities.

External media: editorial authority

Press articles, comparison pieces, “top 10” rankings and specialized media are the main sources behind recommendation answers. Being cited in the most authoritative comparisons in your sector is often the most direct way into an answer.

Forums and review sites: raw reputation

Quora, specialized forums, G2, Trustpilot and Google Business Profile feed the attributes AI associates with your brand: price, reliability, customer service. They often explain why AI describes you as “expensive” or “hard to reach.”

Source

Main engine affected

Effect on the brand

Priority action

Reddit

Perplexity, Google AI Overviews

Tone and social proof

Participate, correct information

YouTube

Google AI Overviews

Demonstration and comparison

Reviews, tutorials, creator coverage

Wikipedia / Wikidata

ChatGPT

Reference facts

Accurate, well-sourced entry

External media

ChatGPT, all

Authority, recommendation

Digital PR

Reviews and forums

All

Perceived attributes

Review collection, public replies

Key takeaway Whatever communities say about you, AI repeats. Our AI visibility platform maps the Reddit threads and YouTube videos AI models cite, from most positive to most negative. It sorts them into four groups: protect, priority, opportunity and watch.

Backlinks: how much do they matter for GEO?

Backlinks still count, but much less than brand mentions for AI visibility. The reference data comes from Ahrefs: web mentions correlate at 0.664 with AI Overview visibility, versus 0.218 for backlinks. The top three correlations are all off-site factors: brand web mentions, branded anchors (0.527) and branded search volume (0.392). ahrefs

The effect compounds. Brands in the top 25% for web mentions average 169 AI Overview mentions, over 10 times more than brands in the next quartile down, which average just 14. At the other end, 26% of brands have zero mentions in AI Overviews. ahrefsahrefs

The picture is similar for ChatGPT. Seer Interactive found weak correlations between ChatGPT brand visibility and link metrics such as domain rank (0.25) and backlinks (0.10). ahrefs

Why backlinks are still useful

Backlinks no longer drive citation directly, but they support classic SEO, and many AI engines rely on search results to choose which pages to consult. A link from a recognized publication also brings a brand mention, which makes it doubly useful.

Signal

Correlation with AI visibility (Ahrefs)

Role in GEO

Brand web mentions

0.664

Primary signal

Branded anchors

0.527

Strong

Branded search volume

0.392

Moderate

Number of backlinks

0.218

Indirect, via SEO

Key takeaway In GEO, a mention of your brand in a reference publication, even without a link, can be worth more than a backlink without context. Aim for brand citations in the sources AI reads, not just links.

How to identify the sources that influence a brand’s visibility

Identifying the sources that influence your visibility means recording the pages cited in every AI answer in your category. You then compare the ones supporting your brand with the ones supporting your competitors. The gap between the two reveals your opportunities.

The manual method

  1. Build a panel of 25 to 50 prompts covering your category: recommendations, comparisons, alternatives, pricing questions.

  2. Run each prompt several times on ChatGPT, Perplexity, Gemini and Google AI Overviews.

  3. Record every cited URL and classify it by type: your site, competitor sites, media, communities, reviews, other.

  4. Calculate your “source share”: the proportion of citations pointing to pages favorable to your brand.

  5. List the domains that cite your competitors but not you: that’s your “source gap.”

To decide which metrics to track and how often, see our guide to the GEO metrics to track.

The AI response flow VizibAI tracks

The manual method quickly hits its limits beyond a few dozen prompts. The VizibAI platform tracks every AI answer in six steps, from the buyer’s question to action:

Step

What VizibAI does

What you get on sources

1. Scope

Builds a prompt set by topic, language and market, with a bias check

Questions that reflect real demand

2. Monitor

Measures visibility, share of voice, source share and sentiment per model

How your source share changes with every run

3. Revenue

Connects GA4 to measure traffic and revenue from each AI platform

The pages AI assistants send traffic to

4. Compare

Shows who AI recommends instead of you, topic by topic

Source breakdown (your brand, competitors, media, community) and source gap

5. Audit

Checks AI claims against your source of truth

The sources spreading errors

6. Act

Turns every insight into an action ranked by impact

Reddit threads to address, content to create, sources to correct

Key takeaway A source only matters when it’s compared: what you need to know is which domains get your competitors recommended instead of you. VizibAI’s “source gap” lists exactly the prompts where a competitor is cited and you aren’t.

How to improve your presence in the sources LLMs use

Improving your presence in LLM sources means acting on three levels: your owned content, reference sources and communities.

Make your website citable

Publish pages that directly answer the questions in your category: FAQs, comparisons, pricing pages, complete product pages, dated facts and figures. Add structured data (schema.org) and check that AI crawlers can access your pages. Update regularly, because older content gets reused less often.

Secure your reference sources

Check that your Wikipedia entry (if your notability allows one), Wikidata, LinkedIn, Google Business Profile and industry directories say the same thing as your website. Any inconsistency ends up in AI answers. If that’s already happening, learn why ChatGPT says wrong things about your company and how to fix the sources behind it.

Earn media mentions

Target the publications, comparisons and rankings AI already cites in your category. Digital PR, original research and expert commentary generate brand mentions, the signal most correlated with AI visibility.

Take part in communities transparently

Reply to the Reddit threads and Quora questions AI cites, identifying yourself as a brand representative and bringing accurate, useful information. Encourage video reviews and creator coverage on YouTube. Collect customer reviews continuously on your industry’s platforms.

Monitor and adjust

Citation preferences change fast. Rerun your analysis at least monthly, compare your source share with your competitors’ and redirect your efforts toward the domains that matter most.

Lever

Target sources

Indicative time to effect

Citable pages on your site

Your domain

Short

Fixing listings and directories

Listings, Wikidata

Short

Community participation

Reddit, Quora, YouTube

Medium

Digital PR

Media, comparisons

Medium to long

Wikipedia

Encyclopedia

Long, depends on notability

Key takeaway AI doesn’t recommend you because you say you’re the best, but because other trusted sources say so. Your website sets the facts, while media and communities provide the proof.

Conclusion

AI answers don’t come from nowhere: they reflect an identifiable set of sources, specific to each engine. Three lessons to remember:

  1. Each engine has its own sources: Wikipedia and media for ChatGPT, Reddit and YouTube for Google AI Overviews, Reddit for Perplexity.

  2. Mentions matter more than links: your brand’s presence in web conversations weighs roughly three times more than backlinks.

  3. Measure before you act: only analyzing cited sources shows where to focus your efforts.

To find out which sources feed the answers about your brand and your competitors, start with a free AI audit.

Frequently asked questions

Which sources does ChatGPT use most?

ChatGPT favors knowledge bases and established media. According to Profound, Wikipedia accounts for nearly half (47.9%) of citations among its 10 most-cited sources, ahead of Reddit (11.3%) and Forbes (6.8%).

Does Reddit really influence a brand’s visibility in AI?

Yes, especially on Perplexity and Google AI Overviews, where Reddit is the most-cited source. A discussion thread about your brand can be reused in an answer, tone included. You need to track the threads AI cites and bring accurate information to them.

Are backlinks still useful for GEO?

Yes, but indirectly. According to Ahrefs, brand mentions correlate at 0.664 with AI Overview visibility, versus 0.218 for backlinks. Backlinks remain useful for classic SEO, which many AI engines rely on.

How can I find out which sources influence answers about my brand?

Run a prompt panel across several AI engines, record the cited URLs and compare them with your competitors’. A platform like VizibAI automates this analysis and calculates your source share and source gap.

How long does it take to appear in AI sources?

It depends on the lever. A well-structured page on your site or a corrected listing can be picked up within a few weeks. Media and community presence takes several months to build. Monthly tracking lets you measure progress.