ChatGPT and Perplexity build their answers from very different parts of the web. Across 961 websites audited with VisibAI between February and October 2026, ChatGPT most often cited Google and Wikipedia, while Perplexity cited Reddit in 94.5% of the audits where it ran and YouTube in 84.4%. If you want to be named by both, you need to be present in both sets of sources.
TL;DR
- Across all 961 sites, the most cited domains were google.com (31.6% of audits), en.wikipedia.org (28.0%), g2.com (19.4%) and linkedin.com (15.9%).
- Perplexity cited Reddit in 94.5% of audits and YouTube in 84.4%. ChatGPT cited them in 2.6% and 3.6%.
- ChatGPT cited press-release wires such as prnewswire.com in 8.5% of audits. Perplexity almost never did (2.8%).
- Social profiles matter far more to Perplexity: Facebook 56.9% and Instagram 42.2%, against under 1% each for ChatGPT.
- Your own website is the only source on these lists you fully control, so it has to be readable, factual and current.
How did we measure which sources AI assistants cite?
Every VisibAI audit asks AI assistants a set of buyer-style questions about a business's category and location, then records every source the assistant cites in its answers. For this analysis we took the most recent completed audit for each distinct website, which gave 961 sites with at least one citation, audited between February 22 and October 12, 2026.
For each domain we counted the share of audits in which it was cited at least once. A domain cited ten times in one audit still counts once, so the figures show how widespread a source is, not how often a single answer leans on it.
Two caveats matter. The sample is self-selected: these are businesses that chose to check their AI visibility, and it leans towards B2B and software. And the engines were not run equally. ChatGPT ran in 944 of these audits, including free ones. Perplexity ran in 109 paid audits, which use more questions per audit. The platform comparison below is still useful, because the gaps are large, but treat the exact percentages as indicative.
Which websites do AI assistants cite most often?
These are the most cited domains across all 961 sites:
| Domain | Share of audits citing it |
|---|---|
| google.com | 31.6% |
| en.wikipedia.org | 28.0% |
| g2.com | 19.4% |
| linkedin.com | 15.9% |
| trustpilot.com | 14.2% |
| youtube.com | 14.0% |
| reddit.com | 13.6% |
| apps.apple.com | 11.4% |
| cbinsights.com | 10.3% |
| capterra.com | 9.4% |
The pattern is consistent: neutral, structured sources sit at the top. Encyclopedias, review platforms, professional networks and business databases give an assistant facts it can verify and repeat. Brand websites appear too, but no single brand site can show up across hundreds of audits the way a review platform does.
Do ChatGPT and Perplexity cite the same sources?
No, and the difference is the most useful finding in this data. Here is the share of audits citing each domain, split by engine:
| Domain | ChatGPT (944 audits) | Perplexity (109 audits) |
|---|---|---|
| reddit.com | 2.6% | 94.5% |
| youtube.com | 3.6% | 84.4% |
| linkedin.com | 8.5% | 83.5% |
| g2.com | 12.3% | 66.1% |
| en.wikipedia.org | 23.0% | 63.3% |
| facebook.com | 0.6% | 56.9% |
| trustpilot.com | 8.5% | 56.0% |
| instagram.com | 0.8% | 42.2% |
| forbes.com | 3.3% | 40.4% |
| capterra.com | 3.7% | 33.0% |
| google.com | 30.2% | 0.9% |
| prnewswire.com | 8.5% | 2.8% |
Perplexity cites more sources per answer, so its percentages run higher across the board. What matters is the shape of each column. Perplexity leans heavily on community and social platforms: Reddit, YouTube, LinkedIn, Facebook and Instagram. ChatGPT leans on Google results, Wikipedia, review platforms and press releases.
What does this mean for where your brand should appear?
It means a single channel strategy will leave gaps. A brand with a strong Wikipedia entry and good press coverage may do well in ChatGPT and still be invisible in Perplexity if nobody discusses it on Reddit or YouTube. The reverse is also common: an active community brand can win Perplexity answers while ChatGPT names an older competitor with better press.
A practical way to read the two columns:
- For ChatGPT: make sure your business appears clearly in Google results, earn coverage on review platforms in your category, and keep press announcements factual and distributed. A Wikipedia entry helps only if your business genuinely meets its notability rules, so do not force one.
- For Perplexity: take part honestly in the communities where your buyers talk, publish short video explainers, keep your LinkedIn company page complete, and collect reviews on Trustpilot or the platform your category uses. Our guide to getting mentioned on Reddit covers how to do this without getting banned.
Where does your own website fit in?
Your website is the one source on these lists that you control completely, and every other source tends to point back to it. When an assistant finds a review or a Reddit thread about you, it often checks your site to confirm what you do, where you work and what you offer. If the site blocks AI crawlers, hides key facts inside images or scripts, or says nothing specific, the assistant has nothing solid to confirm against.
The fundamentals still apply: allow AI crawlers, state your core facts in plain sentences, add structured data, and publish pages that answer the questions buyers actually ask. Our Q&A content playbook shows how to write those pages, and the free AI Crawler Checker tells you in seconds whether AI bots can reach your site.
How can you find the sources that matter in your category?
The domains above are averages across hundreds of categories. Your category will have its own mix: a law firm will see legal directories, a SaaS company will see G2 and Capterra, a restaurant will see local guides. The fastest way to find yours is to ask the assistants the questions your buyers ask and write down every source they cite. Doing that by hand takes a while; a VisibAI audit runs the questions across several engines and lists the cited sources for you, with the competitors that appear next to them.
Frequently asked questions
Which website does ChatGPT cite most often?
In our data, google.com was the most cited domain in ChatGPT answers, appearing in 30.2% of audits, followed by en.wikipedia.org at 23.0% and g2.com at 12.3%. The mix varies by category, so check your own industry before acting on averages.
Why does Perplexity cite Reddit so much?
Perplexity searches the web for every answer and favours sources with recent, first-hand discussion. Reddit threads contain exactly that: real people comparing options. In our audits Perplexity cited Reddit in 94.5% of the audits where it ran, the highest rate of any domain.
Do social media profiles help AI visibility?
For Perplexity, clearly yes: Facebook appeared in 56.9% of its audits and Instagram in 42.2%. For ChatGPT, rarely: both were cited in under 1% of audits. Social profiles are worth keeping complete, but they cannot replace a clear website and third-party coverage.
Does a press release help a brand get cited by AI?
It can help with ChatGPT, which cited prnewswire.com in 8.5% of audits. Perplexity cited it in only 2.8%. A press release works best when it states verifiable facts that other sources can then repeat.
How often should I check which sources AI cites about my business?
Monthly is enough for most businesses. Sources shift as engines update and as new coverage appears, and a monthly check is frequent enough to see whether new content or coverage is being picked up.
See which sources AI uses in your category
Every category has its own citation map, and the sources that decide your category are rarely the ones you would guess. Run a free AI visibility audit to see which sources AI cites when buyers ask about your market, and which competitors appear next to them.