This article was AI-generated as part of an experimental historical-content project. The date reflects the period being analyzed rather than the date the article was originally written.
The Wikimedia Foundation published a set of numbers today that will be quoted as evidence that Wikipedia is fading. After updating its bot detection, the Foundation says human pageviews to Wikipedia over recent months are down “roughly 8%” compared with the same months in 2024.
For a corporate reader, the tempting conclusion is that the company’s Wikipedia article matters a little less each year. Fewer people read it, so fewer people are influenced by it. I think the Foundation’s own explanation points the other way.
What the Foundation actually reported
The context matters. Around May, the Foundation noticed unusually high traffic that looked human, mostly from Brazil. It updated its bot detection, reclassified data for March through August, and found that much of the unusual May and June traffic came from bots “built to evade detection.” It also cautions that the revised figures need to be interpreted “with care,” because the detection rules differ across periods.
Its explanation for the decline is generative AI and social media, “especially with search engines providing answers directly to searchers, often based on Wikipedia content,” along with younger people looking for information on social video platforms.
The myth: fewer visits means less influence
The same post makes the opposite case about influence. “Almost all large language models (LLMs) train on Wikipedia datasets,” it says, and search engines and social platforms “prioritize its information” when answering their users. People are reading what volunteers wrote “all over the internet, even if they don’t visit wikipedia.org.”
So the reach of an article’s sentences has not shrunk. Their route has changed. The text now arrives as a paraphrase in an AI answer, a fact in a knowledge panel, a line read aloud by an assistant. Pageviews measure one channel of that reach, and probably a shrinking share of it.
That is not reassuring for a company. It means an article can influence more people while being read by fewer of them.
The part companies should worry about
The Foundation names the risk itself: “With fewer visits to Wikipedia, fewer volunteers may grow and enrich the content, and fewer individual donors may support this work.”
Consider how errors in a company article usually get fixed. Someone reads it, notices that something is wrong or out of date, and edits it or raises it on the talk page. Readers are a large part of the error-detection system. If fewer people land on the page, fewer are in a position to notice that the acquisition date is wrong, or that the paragraph on a lawsuit never mentions how it ended.
Meanwhile, the sentence keeps getting carried into answers whose readers cannot easily trace it back. A mistake is seen by more people and noticed by fewer. That is the real change, and a pageview chart hides it.
I have argued before that a neutral-looking Wikipedia sentence works as infrastructure for other systems. Infrastructure with fewer inspectors does not become less important. It becomes riskier.
What follows
Do not deprioritize a Wikipedia article because traffic fell. Put it on a review schedule, precisely because the casual readers who once caught errors are less likely to show up.
Check where its content resurfaces. Ask the major assistants and search features basic questions about the company and see which sentences come back almost word for word. Those are the lines carrying the most weight.
And resist the shortcut. Fewer readers may make a quiet, undisclosed edit less likely to be noticed this month. It does not make it less damaging when it is.
The Foundation is asking platforms to credit Wikipedia and send readers back. Companies have their own reason to want the same thing: a reader who clicks through is a reader who might spot the mistake.