Key takeaways
- From Sep 28 to Oct 4, 2026, social platforms made up 12.2% of Google AI Overviews citations and just 0.46% of ChatGPT citations, a gap of roughly 26x (Promptwatch Data).
- ChatGPT used to cite social more. The earlier Promptwatch report had it at 5.8%. Reddit alone fell from 5.19% to 0.09%.
- YouTube is the strongest social bet for Google audiences (4.28% of AI Overviews citations). Facebook (2.38%) now passes Reddit (2.15%).
- Perplexity sits in the middle at 2.7%, with LinkedIn and Reddit tied at 0.86%. X barely registers anywhere.
- The practical rule: stop treating "AI search" as one channel. Plan social effort per engine.
The headline numbers
Here is the claim in one sentence. In Promptwatch's Social Media Citations by AI Model, October 2026 report, covering Sep 28 to Oct 4, 2026, the seven tracked social platforms (Reddit, YouTube, LinkedIn, Facebook, Instagram, TikTok and X) earned 12.2% of Google AI Overviews citations. For ChatGPT the same seven platforms earned 0.46%.
That is not a rounding difference. Two products that people lump together as "AI search" are reading very different parts of the internet.
The data comes from the real interfaces of ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, not from API calls, which matters because user-facing answers and citations can differ from API output. Promptwatch collected it as part of a dataset of 26B+ data points.

Share of citations by engine
| Engine | Social share of citations | Notes |
|---|---|---|
| Google AI Overviews | 12.2% | About 26x ChatGPT. Only engine where TikTok shows up |
| Perplexity | 2.7% | LinkedIn and Reddit tie at 0.86% |
| ChatGPT | 0.46% | Down from 5.8% in the earlier report |
A third-party study from Tinuiti, summarized on LinkedIn, reports different figures: roughly 13% for AI Overviews, 9% for AI Mode and 3% for Gemini. Different prompt sets and time windows produce different numbers, so I would not try to reconcile them to the decimal. What the two sources agree on is the direction: Google's surfaces lean on social far more than the standalone assistants do.

Platform by platform
Google AI Overviews
AI Overviews is the engine where social presence pays off most. Per Promptwatch Data for Sep 28 to Oct 4, 2026:
- YouTube: 4.28%
- Facebook: 2.38% (more than doubled, now ahead of Reddit)
- Reddit: 2.15%
- Instagram: 1.61% (grew)
- LinkedIn: 1.55% (grew)
- TikTok: 0.22%
Two things stand out to me. First, Facebook overtaking Reddit is a surprise, because most GEO advice of the last year said "get on Reddit." Second, Instagram and TikTok show up at all. If you run a local business, a restaurant, a retailer or anything visual, your public Instagram and Facebook pages are now plausible citation sources in Google's answers.
Reddit in AI Overviews has been eroding slowly rather than collapsing. According to Promptwatch's Reddit citations report, its share eased from 2.37% to 2.10% between the first and last seven days of the Jul 7 to Aug 17 window, an 11.3% relative decline. In AI Mode the decline was bigger, from 2.22% to 1.54%.
ChatGPT
ChatGPT has nearly stopped citing social media. All seven platforms together hold 0.46%, down from 5.8% in the earlier social report. LinkedIn is now its top social platform at roughly 0.3%, and Reddit is down to 0.09%.
The Reddit story is the dramatic one. Promptwatch measured Reddit at a steady 3.83% of ChatGPT citations from Jul 18 to Aug 7, 2026. It slipped on Aug 8, the same day ChatGPT changed its query fanout behavior, and dropped below 1% on Aug 14. The Aug 14 to 17 average was 0.52%, an 86.4% relative drop.
A caveat worth keeping: Promptwatch says the cause is unknown. A change in ChatGPT's source selection is the obvious candidate, but a data-collection issue can't be ruled out, so treat the exact size of the drop as provisional. The related site: operator report shows ChatGPT began using the site: operator at scale on Aug 8, jumping from about 0.4% to about 17% of fanout queries. That is a change in how ChatGPT searches, and it could plausibly affect which sources surface.
So where does ChatGPT's attention go instead? Promptwatch's content type data shows product pages as the most cited type at 22.4% (Sep 28 to Oct 4, 2026), followed by landing pages and listicles. Its domain rank data shows sites with DR 46 to 75 earning almost half of citations in August 2026. ChatGPT wants pages that answer a commercial question directly, not threads and posts.
Perplexity
Perplexity cites social platforms 2.7% of the time. LinkedIn and Reddit tie at 0.86%, and YouTube is at 0.79%, down sharply from 2.67% in the earlier report. It is the one engine where LinkedIn is a first-tier social source, which is useful if you sell B2B.
X
X barely registers on any engine. Its peak is about 0.1% on Perplexity. Promptwatch's advice is blunt and I agree with it: leave X out of your AI search plans. Keep posting there for other reasons if you like, but do not expect citations.
Average across the three engines
| Platform | Average share |
|---|---|
| YouTube | 1.70% |
| 1.03% | |
| 0.90% | |
| 0.81% | |
| 0.56% | |
| TikTok | 0.07% |
| X | 0.04% |
Averages hide the engine split, though. Reddit's 1.03% average is mostly Google (AI Overviews and AI Mode) now. For ChatGPT it is close to zero.
Why would the engines differ so much?
Nobody outside the labs knows for certain, so here is my read, flagged as inference rather than fact.
Google has its own index, its own video platform and decades of signals about forums and social pages. Pulling in a YouTube video or a Facebook page is cheap for it. ChatGPT Search relies on a different retrieval pipeline, and its recent fanout changes suggest it is leaning toward queries that target specific sites and pages with clear answers. Product pages, landing pages and listicles fit that pattern. A thread with forty replies does not.
There is also the question of what a citation is worth. Pew Research Center found that in March 2025 browsing data from 900 U.S. adults, users who saw an AI summary on Google were less likely to click result links, and very rarely clicked the cited sources. So a social citation in AI Overviews builds brand presence in the answer more than it drives visits. That does not make it worthless, but it changes what you should measure.

Other third-party work points the same way on a related theme: ranking and citation are different things. Ahrefs, as cited by CXL, found only 37.1% of AI Overview-cited URLs also ranked in Google's organic top 10 for the same query. CXL's own review of 100 AI Overview citations found 55% of cited passages sat in the first 30% of the source page. Cited content is often not the content that ranks, and where the answer sits on the page matters.

What to do about it
If your audience uses Google
Start with YouTube. It is the largest social source in AI Overviews at 4.28%, and the one with the most room for deliberate work: titles, descriptions, chapters and transcripts that state clear answers early. Then keep your public Facebook page and Instagram profile complete and active. Those two are growing as sources, and many brands neglect them.
Reddit still matters for Google, just less than it did. A steady, honest presence is fine. A campaign is not worth it.
If ChatGPT is your priority
Do not count on Reddit threads to get you in. Put the effort into your own site and the formats ChatGPT actually cites: product pages, landing pages, comparisons, listicles and guides. Promptwatch's half-life data adds that landing pages and docs stay cited longest, while news lost half its pages within 36 days. Evergreen beats timely.
If you sell B2B and care about Perplexity
Publish on LinkedIn as well as Reddit. They tie at 0.86%, and LinkedIn is the only platform that appears in the top spots on both ChatGPT and Perplexity. Promptwatch also keeps a LinkedIn citation page types report that shows which formats get cited.
Skip X
It is under 0.1% everywhere. Spend the time elsewhere.
How to track this for your own brand
Report-level averages are a starting point. Your own prompts will tell you more. A practical process:
- Build a prompt set of 50 to 150 real questions your buyers ask, split by topic.
- Run it across ChatGPT, Google AI Overviews, AI Mode and Perplexity, ideally in the country and language you sell in.
- Classify each citation by source type, and flag social platforms separately.
- Compare your social citation share per engine with the benchmarks above.
- Watch rolling trends, not single days. The ChatGPT Reddit drop shows how fast a share can move, and why a one-day dip tells you little.
Promptwatch has dedicated reports for the Reddit, YouTube, LinkedIn and Instagram posts AI cites, plus a "My channels" filter to see only your own accounts, which most trackers skip. Other options exist if you want to compare, and the directories at bestgeosoftware.com and ai-rank-tools.com list many of them.
| Need | Tools worth a look |
|---|---|
| Social citation tracking plus content and crawler data | Promptwatch |
| Prompt-level monitoring | [tool:peec-ai] and [tool:otterly-ai] |
| Enterprise visibility tracking | [tool:profound] and [tool:scrunch] |
| SEO suite with AI tracking | [tool:semrush] and [tool:ahrefs-brand-radar] |
If you want a quick baseline before committing to any tool, Promptwatch's free AI visibility report scans how ChatGPT, Gemini and Perplexity talk about your brand.
Things to keep in mind
- The numbers cover one week (Sep 28 to Oct 4, 2026) for the October report. Shares shift month to month, so check the report before quoting them.
- Shares are a percentage of each engine's own citations, not a count. A small share on a huge volume can still be many citations.
- The cause of ChatGPT's Reddit drop is unconfirmed, per Promptwatch itself.
- Being cited is not the same as being clicked. Pew's browsing data suggests cited sources get very few clicks from Google AI summaries.
The bottom line
AI search is not one channel. Google's AI surfaces cite social platforms about 26 times more often than ChatGPT does, and the platform mix is different too: YouTube and Facebook for Google, LinkedIn for ChatGPT and Perplexity, and almost nothing for X. Pick the engine your buyers use, match the effort to it, and re-check the data every month. For more on how these numbers move, the Promptwatch Data hub is updated regularly.