How Wikipedia lost its ChatGPT lead in 2026
Wikipedia went from 3.63% of ChatGPT citations in January 2026 to 0.56% in June (month-to-date) as Reddit took first place. What the shift teaches about tracking cited domains.
In January 2026, Wikipedia was the single most cited domain in ChatGPT. By June it was a distant second, cited at about a sixth of the leader's rate. Promptwatch's monthly ChatGPT citation share reports trace the change month by month, and the speed of it is the real story: most of Wikipedia's share vanished in one month.
Every figure below comes from those reports, each covering one calendar month of ChatGPT responses where web search was triggered in the UI. There's no February report in the series, but the March report includes February comparison figures. The June report covers June 1 to 27 only, so its numbers are month-to-date.
January: Wikipedia on top
The January 2026 report (January 1 to 31) had Wikipedia at 3.63% of ChatGPT citations and Reddit at 2.58%. Together the two took 6.2%, while the other tracked domains combined came to about 3.3%. LinkedIn had 0.39%, Forbes 0.33%, and TechRadar 0.31%, and the page notes no other domain reached even 0.4%. The page itself is inconsistent about how many domains it tracked: its description says the top 15, while the body refers to 21.
Promptwatch's "The Most Cited Domains on ChatGPT in January 2026" chart (January 1 to 31, 2026), captured November 14, 2026.
Two sites holding nearly twice as much as everyone else in the chart combined is a concentrated picture. It made a simple strategy look reasonable: be on Wikipedia, be discussed on Reddit, and ChatGPT will find you.
February and March: the gap narrows
The March 2026 report (March 1 to 31) carries February figures in its comparison chart. Wikipedia had 3.8% in February and 3.61% in March. Reddit went from 2.96% to 3.21%. Wikipedia's lead shrank from 0.84 points to 0.40. Nothing dramatic yet, but the lines were converging.
The March report also shows the long tail moving between February and March: arXiv rose from 0.14% to 0.34% and Walmart from 0.10% to 0.31%, both cracking the top six. The page sums up that month with a line worth remembering: ChatGPT's sourcing mix is anything but static.
April: the cliff
The April 2026 report (April 1 to 30) is the break. Wikipedia fell from 3.61% to 1.15%, losing over two-thirds of its share in one month. Reddit climbed to 4.14%, roughly 3.6 times Wikipedia's figure, and became ChatGPT's top source. Walmart and TechRadar tied at 0.71%.
Promptwatch's "ChatGPT Citation Share, March vs April 2026" chart, captured November 14, 2026.
The April page's caveat is the most useful sentence in the whole series: a single model or retrieval update can rewrite which sources ChatGPT trusts almost overnight, so treat citation share as a moving target.
May and June: the gap becomes a gulf
In the May 2026 report (May 1 to 31), Reddit jumped 48% to 6.11% and Wikipedia slipped to 0.72%. Reddit was cited more than eight times as often. TechRadar had 0.48%, arXiv 0.31%, and Reuters 0.25%.
The June 2026 report, month-to-date for June 1 to 27, shows Wikipedia at 0.56%, still second. Reddit fell roughly 40% to 3.71%, but that was still more than six times Wikipedia's share.
The full run
| Month | Wikipedia | Report window | |
|---|---|---|---|
| January | 3.63% | 2.58% | Jan 1 to 31 |
| February | 3.8% | 2.96% | from the March report |
| March | 3.61% | 3.21% | Mar 1 to 31 |
| April | 1.15% | 4.14% | Apr 1 to 30 |
| May | 0.72% | 6.11% | May 1 to 31 |
| June | 0.56% | 3.71% | Jun 1 to 27, month-to-date |
Sources: January, March (which also gives the February figures), April, May, and June.
Reddit's own story didn't end in June either. Promptwatch's Reddit drop report shows its ChatGPT share falling from 3.83% (July 18 to August 7, 2026) to 0.52% (August 14 to 17), a drop the page treats as provisional. We cover that in Reddit in ChatGPT before and after August 14. The leader that replaced Wikipedia lost most of its own share four months later.
Why did it happen?
The reports don't say, and we won't pretend they do. They show when each source rose or fell, not the reason.
Our read is that a change in how ChatGPT retrieves sources, rather than anything Wikipedia did, is the likeliest explanation for a fall that steep in one month. The April page's own caveat about model and retrieval updates points the same way. But "likeliest" is doing real work in that sentence. Without a statement from OpenAI, the cause is unknown.
It also helps to see Wikipedia's position on Google. In Promptwatch's AI Overviews reports, Wikipedia sat between 0.08% and 0.15% of citations across the months the text quotes, including 0.15% in the June 2026 AI Overviews report (June 1 to 27, month-to-date). Wikipedia was never a big source for AI Overviews. Its ChatGPT position was the outlier, and that outlier went away.
The lesson for your own brand
Wikipedia isn't your competitor, but the pattern applies to you. A source that ChatGPT relies on in January can lose two-thirds of its citations by April. If your category's answers lean on one review site, one forum, or one publisher, the same thing can happen to that source, and to your visibility through it.
Three things follow:
- Know which domains your answers cite. Not the global top 20; the sources that show up for your prompts.
- Watch them monthly. A one-month spike or drop isn't a trend, as the May page warns. Several months in the same direction are.
- Don't put everything on one source. A brand whose ChatGPT visibility ran mostly through reference coverage would have felt April far more than one that also had strong pages of its own and a presence on the publishers ChatGPT kept citing.
That last point is our interpretation, not something the reports measured, but it's the practical meaning of a mix that can flip in a month.
How to track your cited domains over time
You need domain-level citation data for your own prompts, kept month over month.
In Promptwatch, All Citations lists every cited domain and URL for your monitors, with competitors' domains highlighted. Citation Trends shows citation rate by domain and by source type over time, so a Wikipedia-style slide in a source your category depends on shows up as a line, not a surprise. The Citations dashboard tab (added November 13) keeps cited domains and top cited URLs on one fixed view, and scheduled PDF reports can send it to your team each month. Page Tracker follows specific URLs, for example your brand's Wikipedia article or a review page that cites you.
Tools with domain-level citation trends, ranked
- Promptwatch. Domain- and URL-level citations, Citation Trends with 22 content types and 20 source types, Page Tracker, and scheduled reports. Daily tracking on every plan, from Essential at $95 a month.
- AthenaHQ. Visibility across many models with a GEO score and an Action Center. Starter is $295 a month. Domain-level citation trends aren't documented in our entry.
- Scrunch AI. Core at $250 a month across four LLMs. Our entry notes weekly data refreshes and doesn't document citation breakdowns at this depth.
- Profound. Citation share and answer-level analytics on custom Enterprise pricing.
- Ahrefs Brand Radar. Covers ChatGPT in its AI Visibility Index from $199 a month, retested monthly, with prompts modeled from keyword data.
To see which domains ChatGPT cites for your prompts this month, and how that differs from last month, start a 7-day free trial of Promptwatch on monthly billing and let a monitor run for a full week before reading Citation Trends.