For two years, the standard advice for showing up in AI answers was "get on Reddit and Wikipedia." New data suggests ChatGPT has moved on. In an Ahrefs dataset reported by Search Engine Journal, Reddit was cited in 19.1% of U.S. ChatGPT answers in March and 2.3% in September. Wikipedia went from 14.1% to 2.9% over the same period. Those citations went to the places where information comes from in the first place, not to other summaries. If your content mostly repackages what someone else already said, that matters.

What Ahrefs actually measured

The numbers come from Ahrefs' query fan-out study, published in October 2026. The citation findings use their own dataset: Brand Radar's ChatGPT citations in the US from March to September 2026, covering 2.4 to 3.8 million answers a month. Ahrefs sums up the Reddit result in one line: ChatGPT's Reddit citations dropped 88% in six months.

One detail makes this more than a random swing. Ahrefs checked whether the shift came from people asking different questions. By comparing about 25,000 identical prompts present in both March and September, every site moved in the same direction. It is not an effect of changed questions, but of how ChatGPT chooses its sources.

It also helps to know what's being counted here. This is the share of *answers* that cite a domain. It isn't the share of all citations, and it isn't clicks. As one analysis points out, Ahrefs counts share of answers while Promptwatch counts share of citations, so the numbers are not directly comparable. That's why other trackers can still rank Reddit near the top while this one shows a steep drop. They're answering different questions.

Who gained, who lost

The pattern is clearest in health. PubMed moved up from 64th to fourth place among the domains ChatGPT most frequently cited in U.S. answers from March to September, while WebMD dropped from 18th to 79th. PubMed indexes the research papers themselves. WebMD writes consumer explainers based on them.

The same thing happened in other categories. The citations Reddit, Wikipedia and YouTube lost went to research bodies, regulators, brands and hands-on testers, while explainer and aggregator sites fell. Ahrefs lists the groups specifically. The winners included government health and research: PubMed, FDA, CDC, MedlinePlus, DailyMed, NIH, NHS, plus brand sites and support pages like Apple, Google Support, and Microsoft Support. The losers included health and finance explainers (WebMD, Healthline, Investopedia) along with syndicators and tabloids such as Yahoo, AOL, and the New York Post.

Some well-known sites fell a long way. Britannica, Investopedia, and the New York Post fell out of the top 200. Ahrefs' own reading: ChatGPT "seems to be prioritizing original sources over sites that summarize them."

Why an AI would skip the summarizer

When ChatGPT searches the web, it doesn't run your prompt as one query. It splits the prompt into several smaller searches (a "fan-out"), collects candidate pages, and then picks a few to cite. We covered how often that search step even happens in ChatGPT skipped web search on 6 of every 7 answers. This post is about what happens when it does search.

Reading a page and citing it are separate steps. An earlier Ahrefs study of 1.4 million prompts found that Reddit accounted for 67.8% of the non-cited URL pool, while its dedicated retrieval category had a citation rate of just 1.93%. Ahrefs' summary was that ChatGPT "is using Reddit extensively to understand topics, gauge consensus, and build context—but it almost never gives Reddit the credit."

So the drop doesn't mean ChatGPT stopped reading Reddit. Discussion threads still shape the answer. The citation, the link a reader can actually click, now tends to go to whichever source looks most authoritative for the specific claim: the study, the regulator's page, the manufacturer's spec sheet, the lab test.

That's our interpretation, not something OpenAI has confirmed. OpenAI hasn't announced a change to its citation policy. But it fits the data, and it fits the obvious incentive. An AI that cites the FDA for a drug question is harder to blame than one that cites a forum thread.

What this means if you publish content

If your site explains things that other sites have already explained, this shift works against you. Every summary of the same facts is now competing with the source of those facts, and the source is winning. A few practical takeaways:

Produce something only you can produce. Hands-on testers were among the winners. In the same data, PubMed, the FDA, and testing sites like RTINGS increased their rank. RTINGS doesn't summarize reviews. It runs its own tests and publishes the measurements. Original data, real test results, internal benchmarks, and first-hand screenshots are things a model can't get from anyone else.

Keep your own official pages sharp. Brand sites and support pages gained ground. If you sell a product, your pricing page, docs, and FAQ are now prime citation targets. Out-of-date or vague official pages may get passed over for a third party's description of your product.

Link to primary sources yourself. If you write explainers, point readers to the original study or announcement instead of to someone else's write-up. It's better for readers, and it puts your page closer to the source.

Match titles to specific questions. Ahrefs' earlier research found that titles and URLs matching these narrower queries had a stronger correlation with citations than pages that only broadly matched the original prompt. A page titled around one specific question has a better chance than a broad "ultimate guide."

The caveats worth keeping

This is one dataset from one company, and there are real limits to it.

First, citations aren't traffic. Search Engine Journal says it plainly: Ahrefs' citation ranks don't measure traffic. A higher citation rank doesn't guarantee more visitors.

Second, it's a snapshot. Ahrefs notes that their fan-out data comes from a snapshot of a single model taken between August and September. They also note that these patterns could change in future updates. The citation data covers US answers only.

Third, other engines behave differently. A multi-engine study by Peec AI found Reddit was the most-cited source across ChatGPT, Google AI Mode, Gemini, Perplexity, and AI Overviews. That analysis came out earlier and measured citations differently, so it doesn't contradict the ChatGPT trend. It does mean you can't assume Perplexity or Google's AI features have made the same shift.

The safest conclusion: if you're betting on being one more summary of facts that exist somewhere else, the bet just got worse. If you publish the facts yourself, it got better.

FAQ

Does ChatGPT still use Reddit at all?

Yes, but it credits Reddit far less often. Reddit appeared in about 1 in 5 US ChatGPT answers in March 2026 and about 1 in 43 by late September, per Ahrefs. Earlier research also showed ChatGPT reads plenty of Reddit content that it never cites, so threads can still shape answers without getting a link.

Should I stop posting on Reddit for AI visibility?

Not necessarily, but don't make it your whole strategy. Reddit can still affect how AI tools understand a topic, and it still ranks high on other AI engines. For ChatGPT citations specifically, the data favors original research, official pages, and hands-on testing over community posts.

Is this permanent?

No one knows. Ahrefs says the patterns could change with future model updates, and OpenAI hasn't publicly explained the shift. Check how often AI tools cite your own pages for your core topics, and recheck every few months instead of treating any single study as final.