Digital 2025: exploring trends in Wikipedia traffic
Wikipedia has been an integral part of the internet for almost a quarter of a century, providing free, democratised access to knowledge.
Since its launch in 2001, the site’s community of volunteers has contributed 64 million distinct articles, offering in-depth information on a huge variety of subjects in more than 300 languages.
As a result, Wikipedia reaches a large, global audience: the site itself reports billions of unique monthly device visits, and third-party tools consistently rank Wikipedia.org amongst the world’s most visited web domains.
However, as part of our ongoing research work for the Global Digital Reports series, we’ve noticed some worrying trends in traffic to Wikipedia over recent months.
Initially, we started exploring these trends purely out of curiosity, but what we’ve discovered may have wide-reaching implications for the digital community as a whole.
So, while it differs from our “usual” DataReportal content, we’ve decided to publish our most salient discoveries here, together with an overview of the approach we used to explore these trends.
Side note: this is a long and detailed article, so you might want to grab a drink and a snack before you read any further.
The information superhighway
Let’s begin with some important context.
Our Digital 2025 April Global Statshot Report reveals that “finding information” remains the primary reason why people use the internet today.
This motivation has long topped the rankings in GWI’s regular research, and the company’s data doesn’t show any sign that “finding information” is losing relevance amongst the world’s connected audiences.
Indeed, the share of GWI survey respondents who cite “finding information” as a primary reason for using the internet has actually increased over the past three years, from 60.2 percent in Q4 2021, to 60.9 percent in the company’s most recent Q4 2024 wave of research.
Admittedly, younger generations now say that “staying in touch with friends and family” is the primary reason why they go online, but even then, that motivation is only just ahead of “finding information”.
And moreover, beyond the explicit act of “finding information”, a quest for facts, knowledge, and news is still clearly evident amongst the other top reasons why Gen Z uses the internet.
So, it’s safe to say that “information” remains one of the most valuable resources that the internet offers its 5.64 billion users.
The online oracle
Information takes many forms of course, and internet users turn to a wide variety of online resources to help them find what they’re looking for.
However, across the world’s most popular web domains, the “brand” that people might most associate with information is Wikipedia.
This self-proclaimed “free encyclopedia” has long been a cornerstone of the internet, providing valuable reference for hundreds of millions people around the world each month.
Inevitably, despite its popularity – or perhaps because of it – the site’s “open source” approach isn’t without its detractors, as (somewhat ironically) this page on Wikipedia explains.
However, the very same issues that some observers cite as “weaknesses” could also be seen as strengths, depending on a reader’s willingness to go “behind the scenes” and explore the edits that have been made to each page over time [if you’ve never done this, I strongly encourage you to start; you can learn how here].
Because of the sheer diversity of content that Wikipedia covers – and the frequency with which the site’s contributors update its content – the site has consistently appeared in the world’s top ten domains for as long as we’ve been publishing these rankings.
For example, Wikipedia places seventh in our latest global ranking of websites using Similarweb data, and the site places fifth in our Semrush ranking.
However, Wikipedia’s relative ranking amongst the top ten has been declining steadily over recent months, and the site has lost billions of visits.
Wikipedia’s traffic is evaporating
According to global data from Similarweb, as recently as March 2022, Wikipedia averaged more than 165 million visits per day to its primary “.org” domain.
Just three years later, in March 2025, Similarweb’s data shows that Wikipedia’s traffic had fallen below 128 million average daily visits.
Relatively speaking, these figures point to a decline of 23 percent in total site traffic over the past three years.
But the figures for absolute change offer an even starker illustration of this decline.
Specifically, Wikipedia has lost more than 1.1 billion visits per month over the past 3 years.
For perspective, the amount of monthly traffic that Wikipedia has lost over the past three years is roughly equivalent to the latest figure for total monthly traffic to Microsoft.com, which was the world’s 29th most visited web domain in March 2025.
And even more strikingly, Similarweb’s data shows that – in March 2025 – Wikipedia.org attracted half a billion fewer site visits than ChatGPT.com [note that the Similarweb data published in our Digital 2025 April Global Statshot Report shows average values for the three-month period between 01 December 2024 and 28 February 2025].
It’s important to stress that Wikipedia still attracted roughly 2½ times as many unique visitors in March 2025 as ChatGPT did.
However, Wikipedia’s unique user numbers have also been falling.
Similarweb’s data shows that average monthly unique visitors have fallen by close to 200 million since March 2022, equating to a relative drop of roughly 18 percent over the past three years.
To put that figure in context, since March 2022, Wikipedia has lost more than the total number of monthly unique visitors to sites such as eBay.com, Walmart.com, and BBC.com.
Various tools concur
Other analytics tools tell a similar story too, albeit with different numbers.
For example, data from Semrush shows that total daily visits to Wikipedia.org have fallen by more than 14 percent over the past three years, from an average of 263 million in March 2022, to just 226 million in March 2025.
And while those absolute traffic figures are quite different to Similarweb’s, the overall trend across both tools is eerily similar, with Sermush’s tools also pointing to a decline of more than 1.1 billion visits per month.
Meanwhile, Semrush’s data shows that unique visitors have fallen by an even greater margin.
On a relative basis, the company’s tools show that Wikipedia users fell by 16.5 percent between March 2022 and 2025, with the site losing 228 million unique monthly visitors over those three years.
However, Wikipedia is not the only prominent website facing similar issues.
Wikipedia is not alone
A Kepios study of the 50 largest information-related websites around the world (as ranked by total monthly web visits) shows that the majority of these sites have seen total monthly traffic fall over the past 3 years.
More specifically, based on Similarweb’s data, 36 of the 50 information-related sites that we studied saw their overall monthly site traffic fall between March 2022 and 2025, meaning that more than 7 in 10 of these domains saw their total traffic decline over the past three years.
And perhaps most worryingly, some of the most reputable – and the most popular – information brands on the internet seem to have been the worst affected.
Data published in Similarweb’s tools reveals the following changes in total monthly site visits for a selection of the world’s most visited news and “reference” brands between March 2022 and March 2025 (rank order based on total site traffic in March 2025):
Wikipedia.org: -22.7 percent (-1.16 billion monthly visits)
Globo.com: -22.0 percent (-229 million monthly visits)
Quora.com: -31.6 percent (-275 million monthly visits)
NYTimes.com: -31.7 percent (-316 million monthly visits)
BBC.com: -27.0 percent (-186 million monthly visits)
CNN.com: -36.7 percent (-298 million monthly visits)
DailyMail.co.uk: -41.3 percent (-184 million monthly visits)
TheGuardian.com: -24.2 percent (-107 million monthly visits)
FoxNews.com: -23.1percent (-92 million monthly visits)
Bild.de: -20.9 percent (-52 million monthly visits)
Detik.com: -25.9 percent (-64 million monthly visits)
AajTak.in: -26.8 percent (-50 million monthly visits)
Repubblica.it: -23.9 percent (-36 million monthly visits)
Reuters.com: -21.7 percent (-31 million monthly visits)
WashingtonPost.com: -42.9 percent (-74 million monthly visits)
For balance, it’s worth highlighting that there are some exceptions to this trend, including:
Reddit.com: +42.0 percent (+1.12 billion monthly visits)
Infobae.com: +27.3 percent (+92 million monthly visits)
IndiaTimes.com: +6.8 percent (+11 million monthly visits)
EconomicTimes.com: +125.3 percent (+82 million monthly visits)
APNews.com: +143.5 percent (+87 million monthly visits)
However, despite this handful of more encouraging results, the overall impact of trends across the top 50 information-related sites was a net loss of 3.9 billion monthly visits between March 2022 and March 2025.
So, the trend of declining web traffic is clearly not restricted to Wikipedia.
However, given the relative size of Wikipedia’s audience, together with the breadth and scale of the content that the site contains, it’s worth taking a closer look at why traffic to Wikipedia might be declining.
Wikipedia’s decline in context
For reassurance, Wikipedia is still one of the most popular sites on the internet today, placing seventh in Similarweb’s global ranking, and fifth in Semrush’s charts.
However, as the only “non Big Tech” entity in these lists (perhaps with the exception of “adult” sites), Wikipedia’s success is a crucial bellwether of the current “state of the internet”.
In particular, the ways in which large digital properties treat Wikipedia offers insight into the potential fate of other “non Big Tech” entities such as corporations, publishers, NGOs, and even governments.
As a result, Wikipedia’s declining traffic may have implications for us all.
But what are the underlying drivers of Wikipedia’s declining traffic?
Well, there could be a few explanations, so let’s explore some plausible hypotheses.
Hypothesis 1: could Wikipedia be losing its appeal?
One possible explanation for Wikipedia’s declining traffic is that fewer people are choosing to use its service.
One of the easiest ways to test this theory is to examine how often people proactively type the Wikipedia URL directly into their browser, which is an important component of the “direct” traffic reported in SEO tools like Similarweb and Semrush.
For initial context, when added together, “direct traffic” and “organic search” have consistently accounted for more than 97 percent of all visits to Wikipedia.org over the past three years.
We’ll return to the role of organic search traffic a bit later, but it’s particularly helpful at this point to note that the volume of “direct” traffic to Wikipedia.org has remained remarkably stable since the start of 2022, hovering around 2.2 billion to 2.3 billion visits per month.
In other words, people are just as likely to type the Wikipedia URL into their web browser today as they were this time three years ago.
But for added perspective, let’s explore trends in the use of Wikipedia’s mobile app.
Interestingly, despite launching in April 2012, this app has struggled to gain any significant momentum, and the latest App Intelligence data from Similarweb suggests that global monthly active users currently sit around the 20 million mark.
That’s not an insignificant number of course, but it’s barely 2 percent of the 909 million unique devices that Similarweb reports visited the platform’s website in March 2025.
Crucially, however, those monthly active app user figures have also remained relatively stable over the past three years, albeit with some minor fluctuations from month to month.
And both these trends suggest that there has not been any discernible decline in proactive use of Wikipedia's properties.
Is Wikipedia “top of mind”?
Having said that, Google Trends data tells something of a different story, with global search volumes for the query “Wikipedia” declining by roughly a third* between the start of March 2022 and the end of March 2025.
[*Side note: Google Trends uses dynamic sampling, and this means that the service may provide different values for exactly the same query if the service is accessed at different times or by different users. Indeed, even if you click on the hyperlink to the same query that we used for this analysis, you may see quite different results and values to those we’ve reported here.]
However, as we’ll see later in this analysis, organic search’s contribution to Wikipedia traffic is a particularly complex issue.
For example, many searches that subsequently result in a click to Wikipedia’s website may not necessarily have been initiated by people looking specifically for Wikipedia itself.
Rather, many of these searches may have resulted from an informational need that may not have brought Wikipedia to searchers’ minds as a potential source of answers.
That in itself would be a problem for Wikipedia of course, and the fact that the Wikipedia “brand” may not be top of mind when people look for answers could have profound implications.
Reduced recall
For added context here, the number of people who say that they visit Wikipedia in response to GWI’s regular survey has fallen markedly over recent years.
In the company’s Q1 2022 wave of research, 27.8 percent of internet users aged 16 and above outside of China (where Wikipedia remains blocked) said they had visited Wikipedia within the past month.
By Q4 2024 – the most recent wave for which data were available at the time of writing – that figure had fallen to 22.4 percent, and had fallen as low as 20.4 percent in Q2 2024.
In other words, even after a recent “recovery”, respondents to GWI’s survey say that they’re roughly 20 percent less likely to visit Wikipedia today than they were at the start of 2022.
These figures don’t measure people’s intent of course, and in fact, we should expect to see this level of decline in GWI’s figures given the decline in actual traffic to Wikipedia that we’ve already examined.
In other words, these figures measure the effect rather than the cause.
And while the data we’ve just seen from Google Trends and GWI are certainly cause for concern, the stability of proactive visits to Wikipedia’s digital properties don’t appear to indicate any significant decline in people’s trust in the Wikipedia brand.
Hypothesis 1 conclusion: Wikipedia may be “slipping” from people’s minds, but the data do not conclusively support the hypothesis that Wikipedia has lost people’s favour.
Hypothesis 2: could changes in overall web use be the cause?
Our next hypothesis is that overall web use may have declined.
And in fact, there’s evidence to support this hypothesis.
For example, Semrush’s data shows that total, cumulative visits to the world’s top 100 web domains fell by close to 7 percent between March 2022 and March 2025.
However, the data also reveal meaningful fluctuations – both up and down – during that time, with total visits to the world’s 100 most visited domains by month as follows:
March 2022: 405.8 billion
March 2023: 419.6 billion (YoY: +3.4 percent)
March 2024: 361.0 billion (YoY: -14.0 percent)
March 2025: 378.8 billion (YoY: +4.9 percent)
But perhaps the most telling indicator in this data is traffic to Google.com, which remains the primary source of outside traffic for most of the world’s websites.
And here’s what Semrush’s data tell us about total visits to Google’s website:
March 2022: 146.6 billion
March 2023: 151.2 billion (YoY: +3.1 percent)
March 2024: 133.1 billion (YoY: -11.9 percent)
March 2025: 142.3 billion (YoY: +6.8 percent)
As you can see, these year-on-year change trends are broadly similar to the patterns that we see in overall web traffic.
However, the decline in total web traffic to Google.com from March 2022 to March 2025 was just 3 percent, which is very different to the 23 percent decline that we saw in traffic to Wikipedia.org over the same period.
Similarweb reports a larger drop in traffic to Google.com though, with data indicating a decline in total site traffic of roughly 6.4 percent between March 2022 and March 2025.
So, our first important finding is that overall web activity appears to be trending down.
Figures vary meaningfully across sites and between analysis tools, so it’s difficult to offer an exact figure for the scale of this decline.
But for perspective, Kepios’s analysis suggests that total web activity (i.e. traffic across the web as a whole) has fallen by somewhere between 5 and 10 percent over the past three years.
That’s an alarming trend in and of itself, but the rate of decline that we see in Wikipedia’s traffic is easily 2 to 3 times greater than this average decline that we see across the web as a whole.
Hypothesis 2 conclusion: while an overall drop in web activity may have contributed to the drop in Wikipedia’s traffic, this does not appear to be the primary cause of Wikipedia’s specific woes.
Hypothesis 3: has search engine use declined?
As we’ll explore in considerably more detail later in this analysis, one of the biggest factors contributing to Wikipedia’s traffic decline has been a decline in referrals from search engines.
So, one of our first questions was whether an overall decline in the use of search engines might be to blame for Wikipedia’s lost traffic.
For broader perspective, much has been made in the media over recent years of “the death of search”, with one of the most cited comments coming directly from a Google exec:
“40 percent of young people – when they’re looking for a place for lunch – they don’t go to Google Maps or Search; they go to TikTok or Instagram.”
However, while there’s certainly truth in the specifics of this particular statement, the media has frequently distorted its message.
For clarity, this quote references an internal Google study of how people aged 18 to 24 in the United States conduct local searches for things like where to eat lunch.
That’s a very specific search behaviour amongst a very specific demographic that accounts for barely 0.5 percent of the world’s internet users.
And much as this soundbite offers fascinating insight into people’s evolving digital behaviours, a change in the restaurant search behaviours of Gen Z Americans is not – in and of itself – indicative of any potential “death of search”.
Indeed, the latest research from GWI shows that the use of search engines has remained relatively stable over recent years, with more than 4 in 5 online adults (81.5 percent) still using search engines every month.
And moreover – in stark contrast to those clickbait headlines you may have seen in the media – younger demographics are actually amongst the most prolific users of search engines.
Admittedly, traffic to Google.com has declined over recent years, but as we concluded in the previous section, this trend is more likely the result of a general decline in web usage, rather than a reduction in the use of search engines specifically.
And for one last perspective, data from both Semrush and Similarweb also indicates that traffic to the world’s second most popular search engine – Bing.com – has actually increased meaningfully over the past three years.
Hypothesis 3 conclusion: the available data do not support the hypothesis that changes in people’s use of search engines is a primary cause of the decline in Wikipedia’s web traffic.
Hypothesis 4: have changing device habits played a role?
Another interesting hypothesis is that changes in people’s device preferences may have resulted in changes in the ways people consume information online, with particular reference to the steady rise of mobile.
For initial context, mobile phones are by far the most popular devices used for accessing the internet today, with roughly 96 percent of online adults saying that they use a mobile handset to go online today.
Meanwhile, figures from data.ai that we published in our Digital 2025 Global Overview Report showed that web browsers account for less than 6 percent of the world’s total mobile time.
In other words, mobile users are significantly more likely to use native mobile apps than they are to visit websites.
That finding may be problematic for Wikipedia, because – as we saw earlier – its app users only account for a tiny fraction (≈2 percent) of the platform’s overall user base.
However, the imbalance between app and web use isn’t a primary consideration for our current study, because our findings focus on changes in traffic to Wikipedia’s website.
But despite web browsers’ slim share of overall mobile time, data from Statcounter indicates that mobile phones still accounted for almost two-thirds – 62.2 percent – of global web traffic in February 2025, up from 56 percent in February 2022.
And for clarity, that 6 percentage-point change equates to a relative swing of more than 11 percent in mobile’s favour between February 2022 and February 2025.
However, while these trends are certainly interesting, our analysis suggests they’re not a key factor in Wikipedia’s traffic declines.
Indeed, Semrush data shows that mobile’s share of traffic to Wikipedia.org has remained relatively stable over the past 3 years, hovering between 23 and 24 percent.
Hypothesis 4 conclusion: while evolving device preferences are certainly important trends, it does not appear as though the “move to mobile” has been a primary factor in Wikipedia’s traffic declines either.
Hypothesis 5: has Wikipedia seen diminished search performance?
If we explore the timing of changes in Wikipedia’s traffic, a particularly interesting hypothesis emerges.
For context, data from Semrush indicates that total visits to Wikipedia.org in March 2022 were broadly the same as the traffic that the site enjoyed 6 years earlier, at the start of 2018.
It’s important to highlight that there were some “ups and downs” in Wikipedia’s web traffic in the interim, and Semrush also changed its methodologies during this time, so this may not be a truly “like-for-like” comparison.
However, we see Wikipedia’s traffic begin a more consistent decline from the first quarter of 2022, and similar patterns are clearly evident in data from both Similarweb and Semrush.
And critically, the timing of this change aligns exactly with a steady decline in the share of Wikipedia’s traffic that comes from organic search.
Hypothesis 4 conclusion: the correlation in timing and the similar scale of associated declines suggests that this hypothesis is worthy of further investigation.
Organic decay
Despite Wikipedia’s sustained popularity amongst web users – as evidenced by the stability of its direct site traffic – search engines are now significantly less likely to send traffic to Wikipedia than they were at the start of 2022.
Specifically, Semrush reports that organic search traffic to Wikipedia has fallen from roughly 5.8 billion site visits in January 2022, to just 4.3 billion in March 2025.
In other words, Semrush’s data suggests that search engines send 1.5 billion fewer visits to Wikipedia.org today than they did at the start of 2022.
On a relative basis, these figures reveal a 26 percent decline in organic search traffic to Wikipedia.org over the past 3 years.
And once again, while the absolute numbers differ, Similarweb’s data tells a very similar story, revealing a 23 percent decline in organic search traffic over the past 3 years.
It’s worth highlighting that Similarweb’s data indicates a gentle decline in Wikipedia’s direct traffic too though, with figures for March 2025 roughly 12 percent lower than the equivalent figures for March 2022.
However, the absolute size of these changes is of greater significance, because organic search accounts for a far greater share of traffic in Similarweb’s data.
More specifically, Similarweb’s data shows a drop of roughly 47 million direct visits per month between March 2022 and March 2025, from 373 million to 326 million.
However, organic search visits fell by more than 1.08 billion during the same period, from 4.61 billion in March 2022, to 3.53 billion in March 2025.
In other words, while both direct and organic search visits fell over the past three years, Similarweb’s data suggests that the change in organic search referrals was responsible for 96 percent of the total decline in Wikipedia traffic over the past three years.
Interpreting Wikipedia’s traffic trends
The clear takeaway here is that search engines are delivering far fewer clicks to Wikipedia today than they did just three years ago.
And with data from Statcounter highlighting the fact that Google has been responsible for somewhere between 89 and 93 percent of all global search clicks over the past 3 years, what this data really tells us is that Google is delivering far fewer clicks to Wikipedia than it used to.
But why?
If “macrodigital” trends aren’t the cause of recent traffic changes, could those changes perhaps be due to other sites or information sources taking traffic away from Wikipedia?
Hypothesis 6: is ChatGPT the culprit?
And the most obvious question here is whether services like ChatGPT might have stolen Wikipedia’s glory.
Somewhat surprisingly, that question doesn’t have a straightforward answer.
However, the most important takeaway from our analysis is that – despite its impressive ascent – ChatGPT doesn’t appear to be the primary culprit of declines in Wikipedia’s traffic.
As we saw above, most of Wikipedia’s recent declines appear to be the result of a decline in organic search referrals, and – at the time of writing – search engines like Google do not surface links to third-party LLMs like ChatGPT, Perplexity, or Claude in response to informational (“know”) search queries.
As a result, it’s unlikely that these LLMs have stolen Wikipedia’s organic search referral traffic.
Having said that, if people are increasingly turning to LLMs like ChatGPT instead of using search engines like Google for informational queries, that could potentially have played a part in these declines.
This is certainly a valid hypothesis, and – as we’ll see later – the rise of services like AI overviews has very likely contributed to Wikipedia’s decline.
However, data from both Semrush and Similarweb suggests that more than half of recent reductions in Wikipedia’s organic search traffic occurred before ChatGPT launched to the public at the end of November 2022.
Moreover, as we saw earlier, research from GWI suggests that search engines are just as popular as they’ve ever been.
And furthermore, while traffic to Google.com has fallen by roughly 3 to 6 percent over the past 3 years, we’ve already seen that this finding on its own isn’t sufficient to explain the scale of declines we’ve seen in Wikipedia’s traffic.
In other words, it appears as though the rise of LLMs and standalone AI tools like ChatGPT has only exacerbated a trend that was already well in play by the time LLMs burst onto the scene 2½ years ago.
Hypothesis 6 conclusion: it does not appear as though standalone LLMs have been the primary cause of declines in Wikipedia’s traffic.
Hypothesis 7: does Wikipedia face a different kind of competition?
So could traffic declines have been caused by a different kind of competitor?
That’s a tricky question to answer, because – as a non-profit and an “open source” encyclopedia – Wikipedia enjoys a somewhat unique position amongst its peer set.
But for perspective, other “reference sites” have experienced similar – and often more dramatic – declines in traffic.
For example, Britannica.com has seen its total traffic fall from 69.5 million visits in March 2022, to just 47.4 million visits in March 2025, for a relative decline of 32 percent.
Meanwhile, Reference.com has seen its traffic practically disappear over the past three years, from 10.7 million monthly visits in March 2022, to just 730,000 monthly visits today.
That’s a 93 percent decline in just three years, and Similarweb’s data suggests that the site’s traffic has essentially flatlined for the past 12 months.
Similarly, traffic to encyclopedia.com has fallen by more than 77 percent in the same period, from 4 million monthly visits in March 2022, to fewer than 1 million monthly visits today.
Looking more broadly, Quora.com has also seen its traffic fall by more than 30 percent, from 870 million monthly visits in March 2022, to just 595 million in March 2035.
However, almost all of that decline in Quora traffic has taken place over the past twelve months, and in fact, total visits to Quora.com were close to 1 billion as recently as March 2024.
So, it seems as though Wikipedia is not alone amongst its reference set in experiencing sizeable declines in web traffic.
But there’s one critical exception to this trend: Reddit.com.
Reddit rising
Reddit has been gaining significant momentum over recent months.
The platform has risen quickly through the ranks of the world’s top websites, reaching ninth position in our latest Similarweb chart, and sixth place in our Semrush ranking.
We’ll leave it to you to decide whether Reddit and Wikipedia are “similar” in terms of their respective content and user experiences, but there’s no denying that Reddit is an increasingly important – and increasingly popular – source of information on the web.
Indeed, our analysis of data from both Semrush and Similarweb reveals a significant increase in traffic to Reddit.com over the past three years.
Once again, these two tools offer slightly different figures for site traffic, but both datasets suggest that total traffic to Reddit’s website has increased by more than 1 billion monthly visits since the start of 2022.
And strikingly, across both datasets, the absolute declines that we see in Wikipedia’s traffic over the past three years are remarkably similar to the absolute increases that we see in traffic to Reddit.
Change in total monthly site visits over the past 3 years: Similarweb data
Wikipedia: -1.16 billion
Reddit: +1.12 billion
Change in total monthly site visits over the past 3 years: Semrush data
Wikipedia: -1.12 billion
Reddit: +1.43 billion
Reddit’s search for success
However, given these figures, it’s interesting to note that “direct” traffic to Reddit.com has remained relatively consistent during that same period.
Naturally, there has been some variation in monthly direct visits to Reddit’s website over the past three years, and Semrush’s data in particular shows an elevated volume of direct traffic for much of 2022.
However, the key finding in data from both Similarweb and Semrush is that total direct visits to Reddit.com were actually slightly lower in March 2025 than they were in March 2022
In other words, both datasets suggest that people are now less likely to type Reddit’s URL directly into their browser’s address bar than they were this time three years ago.
And all of this data suggests that Reddit’s dramatic rise over the past few years can be attributed almost entirely to improved organic search performance.
Google heart Reddit?
So why have search engines – and more specifically, why has Google – been sending so much more traffic to Reddit over recent months, when so many other sites have seen their traffic decline?
Well, this is a topic that has already gained significant attention in the SEO community, with numerous articles on the subject appearing across top industry sites (some interesting examples here, here, and here).
Moreover, Google itself has acknowledged that – to a certain extent – it now “prioritises” Reddit in search results, with the company’s “Search Liaison” account on X offering the following explanation:
“[A]ctual searchers seem to like [forum content]. They proactively seek it out. It makes sense for us to be showing it to keep the search results relevant and satisfying for everyone.”
Meanwhile, an article on Google’s blog adds additional context (emphasis mine):
“In addition to making it easier to find authentic perspectives, we’re also improving how we rank results in Search overall, with a greater focus on content with unique expertise and experience. Last year, we launched the Helpful Content System to show more content made for people, and less content made to attract clicks. In the coming months, we’ll roll out an update to this system that more deeply understands content created from a personal or expert point of view, allowing us to rank more of this useful information on Search.”
For added perspective, it may be helpful to reference this short overview of Google’s Helpful Content System, with particular reference to the timing of the initial announcement (emphasis mine):
“Announced in 2022 as the “Helpful Content Update”, this was a system designed to better ensure people see original, helpful content written by people, for people, in search results, rather than content made primarily to gain search engine traffic. In March 2024, it evolved and became part of our core ranking systems, as our systems use a variety of signals and systems to present helpful results to users.”
So, since 2022, it would seem that Google has been proactively “prioritising” content written by humans, for humans, especially if that content appears to have a “personal or expert point of view”
And that timing coincides neatly with the rise we’ve seen in Reddit’s organic traffic, although it’s worth highlighting that Reddit had already started to gain organic search momentum shortly before this Google update was announced.
But these explanations beg another question.
If Google does indeed prioritise Reddit because the site contains content written by humans with “unique expertise and experience”, why wouldn’t that apply in equal measure to sites like Wikipedia and Quora too?
Arguably, both of these sites offer similar kinds of “community generated content” to what we see on Reddit, yet both sites have seen significant declines in their organic search traffic over recent months.
Affinity vs. prompted action
Part of the reason may lie in how people use Google to search for information.
Analysis of Google Trends data reveals that the volume of Google searches that include the term “Reddit” has more than doubled over the past 3 years.
Meanwhile, at a worldwide level, queries overtly mentioning “Wikipedia” have declined by a third, while queries including “Quora” have declined by close to 60 percent during the same period.
These figures in themselves might be enough to explain why Reddit has seen such an impressive rise in organic search referrals over the past three years, at the same time as Wikipedia and Quora have seen their organic search traffic decline.
However, it’s important to highlight that Google has been adding “Reddit” to original search queries as one of its suggested “People also search for” (PASF) results for at least a couple of years.
And if a user clicks on one of these “suggested” searches, Google then runs the PASF query as an entirely new search, meaning that each PASF search actively contributes to Reddit’s apparent search volume in Google Trends data.
As a result, it’s unclear whether searchers have been proactively entering Reddit more frequently as a query in their own searches, or whether Google’s “people also search for” triggers are the primary driver of Reddit’s increased query volume in Google Trends data.
But why might Google be prioritising Reddit, while Wikipedia and Quora appear to be losing out?
Deal or die
Interestingly, there’s nothing apparent in publicly available data that offers a conclusive answer to that question.
However, it’s worth noting that Google has signed a commercial deal with Reddit, and the timing of that partnership coincides neatly with a significant rise in the Google rankings of tens of millions of pages across the Reddit platform.
As SEO expert Glenn Gabe of G-Squared reported on X, the number of Reddit page URLs that rank in the top 100 Google results for at least one search query leapt by almost 100 percent in the space of just a few months, from roughly 22 million in May 2023, to 41 million in January 2024.
In Glenn’s own words,
“[T]hat’s a massive jump... it’s not just rankings that improved, it’s the pure number of pages ranking.”
For context, the cited partnership involves Google paying Reddit USD $60 million per year, in return for Google being able to use Reddit’s content and API to train its AI tools.
As Google stated in its announcement of this partnership (emphasis mine):
“Over the years, we’ve seen that people increasingly use Google to search for helpful content on Reddit to find product recommendations, travel advice and much more. We know people find this information useful, so we’re developing ways to make it even easier to access across Google products.”
Those final words – “making [content on Reddit] even easier to access across Google products” – certainly seem to imply that Google has been actively and specifically amplifying the visibility of Reddit content across its products, potentially including Search.
But the following sentence from the same statement may be even more revealing (again, emphasis mine):
“This partnership will facilitate more content-forward displays of Reddit information that will make our products more helpful for our users and make it easier to participate in Reddit communities and conversations.”
More specifically, the second half of that sentence may explain why Google has been so active in promoting Reddit content.
Now, given the Google Trends data we saw earlier, it’s important to acknowledge that significant numbers of people probably do use Google to actively search for information and recommendations on Reddit.
However, Google will clearly benefit from an expanded and enriched AI training dataset if more people contribute more content to Reddit.
So, in light of this partnership, there seems to be a huge – and incredibly valuable – incentive for Google to send a greater volume of searches to Reddit.
Interestingly, the same statement goes on to say the following (emphasis still mine):
“This expanded partnership does not change Google’s use of publicly available, crawlable content for indexing, training, or display in Google products.”
However, my reading of this is paragraph is that Google will simply continue to crawl Reddit as it has always done, in addition to using the new information that it receives via Reddit’s API to train its AI systems.
But critically, this sentence does not explicitly say that the partnership between the two companies won’t result in Google promoting Reddit more actively in its search results.
In many ways, this wouldn’t be all that dissimilar to the improved visibility that Google offers its paid search customers.
However, the key difference is whether Google’s users are alerted to the fact that certain content has gained advantageous visibility through a commercial partnership, and whether knowledge of such a partnership would change searchers’ subsequent behaviour.
Contextual aside: Google’s commercial imperative
At this point, it’s critically important to remember that Google is a commercial enterprise – not a public utility.
Google needs to make money in order to survive, and to continue providing the various services on which its billions of users have come to depend.
Of course, the company’s financial results suggest that Google has been very successful at making money, but in some cases, organic search activities incur costs that actively reduce the company’s bottom line.
For example, every single user search requires electricity to retrieve and process relevant answers, and then to deliver those answers to the user.
Moreover, Google’s backend systems must constantly crawl the web in order to collect the signals that feed its search results, and those systems must also categorise and process this data once it’s been collected.
As such, every user search costs Google money.
Those costs are justified business expenses of course; Google’s paid search products generate hundreds of billions of dollars per year, and without its organic results, that paid business would not exist.
Moreover, clicks on organic search results can still create value for Google if a destination site shows Google Display ads, or if it hosts and delivers its content using Google Cloud infrastructure.
However, Google knows that AI poses an existential threat to its Search business, and the company must plan and prepare for a future where Search may not deliver the core of the company’s revenues.
So, while other organisations, governments, and end users may not like the realities that such preparations incur – such as paid content partnerships – it would be naive to expect Google to prioritise the success of any other organisation over its own.
Yes, we should always expect companies like Google to uphold the highest legal and ethical standards.
And yes, there should be severe consequences should Google ever actively discriminate against any legitimate source or destination.
However, the expectation that Google should freely and equally support every site on the internet is unrealistic; by their very definition, search rankings must prioritise one answer over another.
Moreover, unless we turn Google into a public utility, we can only expect the company to actively support those companies and organisations that help ensure Google’s own survival.
Reddit vs. Wikipedia
But returning to our data, could Google’s “amplification” of Reddit content in its search results really explain the dramatic declines that we’ve seen in Wikipedia’s traffic over the past 3 years?
And in particular, could content on Reddit really answer the same questions that content on Wikipedia might otherwise answer?
Our initial hypothesis was that the two sites offer quite different kinds of content, and we were sceptical of the idea that Reddit might be able to steal that volume of traffic from Wikipedia.
But then we realised that the 1.1 billion monthly visits that Wikipedia has lost over the past three years “only” accounted for around a quarter of the site’s traffic.
And if we change the question to ask whether content on Reddit might be able to answer 1 in 4 of the questions that content on Wikipedia might otherwise answer, that seems more plausible.
Indeed, Similarweb reports that there is a 64.2 percent “similarity” between Wikipedia.org and Reddit.com, where “similarity” refers to parallels in content, audience, keywords, and – crucially – in third-party referrals.
And for added context here, it may be useful to highlight that links and referrals are some of the primary “signals” that inform a page’s eventual ranking in Google’s search results:
As Google’s own resources state:
“[W]hen links are merit-based and volunteered as an editorial choice, they’re one of the positive signals to Google about your site’s importance… [R]elevant, quality inbound links can affect your PageRank.”
And given that both Wikipedia and Reddit are such huge sites, there’s a good chance that they both have similarly strong scores for related content, even if there’s a strong likelihood that such “similar” content will be presented in very different ways across the two sites.
So, yes: it’s certainly feasible that content on Reddit could indeed answer 1 in 4 of the questions that might otherwise have been answered by content on Wikipedia.
Hypothesis 7 conclusion: given all of the evidence we’ve explored, it seems quite possible that Google’s “amplified visibility” of Reddit content within its search results could have played a meaningful role in impacting Wikipedia’s traffic.
Initial conclusions
When we began this investigation, we suspected that AI might be the primary cause of declines in Wikipedia’s web traffic.
In the end, those suspicions appear to have been borne out, but not in the ways we expected.
More specifically, we wondered whether people’s use of LLMs and AI overviews might have started to replace the use of search engines and reference websites.
That doesn’t appear to have been a primary contributing factor though – at least not yet.
Rather, it appears as though AI companies’ need to train their models has already delivered a meaningful part of the impact that we expected to see later, due to changes in human behaviour.
And this has huge implications for anyone working in digital, because it means that we may not need to wait for our audience’s behaviours to change before we experience the full force of AI disruption.
But that’s not all…
However, the increased prominence of Reddit content within Google’s search results may not be the only challenge that Wikipedia needs to contend with over the coming months.
Zero-click results
Google’s search results pages have evolved significantly from the simple list of “10 blue links” that characterised the platform’s early days.
Features such as snippets, “people also ask” suggestions, knowledge panels, and – increasingly – AI overviews now occupy a meaningful share of the real estate on the first page of Google results.
Indeed, it’s now not uncommon for searchers to have to scroll beyond the bottom of the first screen before they find the first link to an external (non-Google) site, especially if their search involves a straightforward, “informational” query.
For example, at the time of writing, a factual query like “capital of Scotland” on a desktop browser generates a results page that includes:
An “instant answer”, showing “Edinburgh” in relatively large text at the top left-hand side of the page.
A “People also ask” box immediately below that, wherein a click will generate a new Google search using the suggested query.
A prominent “knowledge panel” on the right-hand side of the screen that presents a short overview of the city, together with a handful of key facts. The source of this content is cited as “Google”, and each of the hyperlinks within the knowledge panel trigger a new Google search for the given query.
A list of videos that all take the searcher to YouTube.
And even at 1920x1080 screen resolution, this is all that’s visible on the initial results “page”.
And critically, users need to scroll down beyond the bottom of that initial screen before they reach any links to third-party content (although, admittedly, the first hyperlink “below the fold” is currently to a Wikipedia page).
But here’s the thing: many searchers will already have found what they need in the content displayed “above the fold”, and won’t need to click on anything else – hence the term “zero-click results”.
This might be frustrating for publishers and for marketers, but we need to recognise that Google’s priority is to keep searchers happy, and to ensure that those users come back next – and every – time they have a question.
Moreover, to put it bluntly, most searchers are looking for answers, not hyperlinks.
And furthermore, in many cases, searchers don’t really care about the source of the answer, as long as it meets their needs.
You may understandably find that troubling, but we’ve seen ample evidence that suggests convenience and “an experience” often win out over the truth, the whole truth, and nothing but the truth.
These stark realities have important consequences for sites like Wikipedia, especially as AI overviews “resynthesise” more complex content that they have retrieved from third-party domains.
Indeed, in the near term, the rise of these “zero-click” search results – and especially the prominence of Google’s own AI overviews – may further reduce the number of searches that result in a click through to Wikipedia’s website.
For context, a recent study conducted by the team at Ahrefs found that search results that include an AI overview result in a third fewer clicks to external sites compared with search results that don’t include AI features:
“We analysed 300,000 keywords and found that the presence of an AI Overview in the search results correlated with a 34.5% lower average clickthrough rate (CTR) for the top-ranking page, compared to similar informational keywords without an AI Overview.”
And if you’d like to dig deeper into the impact of “zero-click” results, you may want to check out this article from SEO legend, Rand Fishkin.
But worse may be yet to come.
The AI dilemma
Recent developments suggest that Wikipedia and its peers may face even greater existential threats than those posed by zero-click search results and commercial content partnerships.
In the longer term, LLMs like ChatGPT, Perplexity, and Claude could easily become primary destinations for the kinds of activities that currently go to search engines like Google and reference sites like Wikipedia.
And what’s more, this “longer” term may not be all that distant.
For clarity, based on everything I see in the data – and more than 25 years of studying digital trends – I fully expect that Google and Wikipedia will remain central pillars of the online world for many years to come.
However, the data also show how quickly AI platforms have been gaining momentum.
For example, Kepios analysis of data from a variety of reputable sources indicates that more than half a billion people already use LLMs each month, and that figure is 2½ times higher than it was just twelve months ago
And while it may still take a few more years, current trends suggest that AI may eventually become the world’s primary source of knowledge and answers.
But therein lies a fundamental problem.
Despite incredible advances in the sophistication of LLMs over the past few years, and despite the promise of even greater innovation to come, these tools are still dependent on “inputs”.
And while I have little doubt that every word of Wikipedia’s 64 million articles has already been fed into these tools’ underlying models, a meaningful proportion of that content still needs to be updated on a regular basis in order for it to remain valid and valuable.
However, there’s no practical way for today’s AI tools to collect and synthesise the “inputs” that would be required to make these updates entirely on their own.
Critically, AI tools are still largely dependent on humans to collect and publish the “raw” content they need, before they can make use of this information within their underlying models.
For example, ChatGPT will happily tell you how many social media users there are in the world, but the platform’s models still rely on third-party sources for their answers, and – in most cases – the quoted figures ultimately rely on DataReportal’s original research [if you’d like to know the answer, click here].
This reliance on human-generated inputs may change over time of course, but we’re still a long way away from a scenario in which legions of autonomous AI agents roam the physical world and the wider universe collecting primary research information.
Indeed, that’s precisely why Google has signed that $60 million annual deal with Reddit: AI is a beast that needs constant feeding, and it has a voracious appetite.
But this ongoing need for inputs creates a troubling catch-22.
As AI tools increasingly take traffic and attention away from services like Wikipedia and news sites, those services are increasingly starved of the revenue that they need to survive.
And if they don’t survive, they will no longer be able to provide the inputs that those very same AI systems also need to survive.
Even in Wikipedia’s case, if fewer people visit the site, we might expect fewer volunteers to contribute to its content, and fewer visitors will perceive an incentive to donate money to support the site’s upkeep.
In other words, AI isn’t just biting the hand that feeds it; it’s actively devouring the entire body.
And given current trends, there’s a real danger that many of today’s most valuable sources of information will become financially unviable long before they’ve been replaced by autonomous AI agents.
For clarity, I’m a firm believer in the power and potential of LLMs and broader AI.
However, I also believe that we need balance; ultimately, we can only reap what we sow.
And for the foreseeable future, that means we must ensure the survival of sites like Wikipedia.
Essential takeaways
So what does all of this tell us?
Here are just a few of the things we’ve been thinking about as we try to make sense of these trends:
Value what’s valuable: the AI industry currently does very little to reward the people and the organisations that create the information and content upon which it depends. Irrespective of debates around copyright, this is simply unsustainable.
Prioritise outcomes, not outputs: in many cases, content and delivery channels are simply means to an end. How else can we achieve the things that matter to us – as individuals, as organisations, and as societies – beyond what we do today? For example, how else might Wikipedia “democratise knowledge” if its website no longer delivered that goal?
Visibility and access are essential: even the best content, products, and services will fail if they aren’t easy to find. Advertising and marketing can be powerful remedies here, but partnerships may play an increasingly important role too.
Convenience often wins: in many instances, “good enough” is all people need, and the first answer often suffices.
Progress is merciless: even the best loved brands may die if they fail to innovate and remain relevant in new realities.
Diversification is essential: this can be tricky when our audiences have become so heavily dependent on a single source of answers, but we must do all we can to find multiple ways to reach and engage our audiences.
Actively manage IP: in addition to protecting things like copyright, organisations may want to take more proactive steps to create value from their IP, especially in new contexts and or unconventional environments.
Be wary of “free”: it can be risky to depend too heavily on something that a third party currently provides for free, because – sooner or later – somebody always has to pay.
What did we miss? I’d love to hear what you think, so please do share your thoughts with me over on LinkedIn.
Click here to see all of Simon’s articles, read his bio, and connect with him on social media.