By For 25 years, web search meant Google (and a little bit of Bing, Yahoo, and DuckDuckGo). But in the past 2.5 years, it’s often expanded to include the major AI tools: ChatGPT, Claude, Deepseek, CoPilot, and Gemini. We found this broader definition welcome, but a little strange. If ChatGPT is search, why isn’t Instagram? Or YouTube? Or Amazon? It struck us, and our clickstream data partners at Datos (a Semrush company), that search should be inclusive of ALL the places where humans on the Internet conduct queries to find information, products, services, and answers. For the last two years, we’ve been jointly releasing our quarterly State of Search reports (which are inclusive of traditional, AI, and e-commerce search), but wanted to go even broader.
Our Theory: search isn’t just traditional engines and AI tools – it’s happening everywhere from social networks to content platforms to e-commerce sites, and beyond. In this research, thanks to Datos’ extensive desktop device panel, we explore the quantity of search volume that happens across the web.
Methodology: we used Datos’ 2025 desktop panel of millions of devices in the United States and the 27 EU member countries plus the United Kingdom, broken down by month/quarter to provide the data below. We analyzed 41 domains for this study across numerous verticals (traditional search, e-commerce, AI tools, reference sites, travel, real estate, classifieds, and more). These were editorially chosen by Rand, based on the most visited 250 domains in Datos’ panel, narrowed down by those with relevant search behavior (e.g. yes to Instagram, no to USPS.com). Not every site we wanted to analyze was part of this, and we hope to do even more in the future, but it’s the most comprehensive analysis of its kind by far.
Conclusions: I guess you could jump right to ’em if you want to miss all these lovely graphs 😉
14 Websites Reach 20%+ of Americans; 15 Reach 20% of EU/UK
We started with an analysis of panelists visiting the 41 domains, and their 2025 traffic trend.
This isn’t search (yet, we’ll get to that), it’s the percent of the millions of devices that make up Datos’ panel who visited each domain at least 1X each month of the quarter, shown over Q1-Q4 2025. The most notable trends for me was the decline for Wikipedia (likely from Google sending them far less traffic than they once did), the rise of Instagram’s desktop traffic (in years prior, they were mobile-only, and then had limited desktop features, but they’ve clearly become a force on desktop, too), and perhaps most interestingly, the growth (then flattening) of ChatGPT.
We looked more closely at ChatGPT’s trend in the 2025 State of Search report, and I’ve pasted that graph below for reference:
There’s been a narrative in many tech and pop culture circles that AI (especially ChatGPT) will destroy/dominate/surpass Google. And yet, even on desktop (where ChatGPT should be strongest), it looks as though they’ll top out behind Reddit, Amazon, Facebook, YouTube, and of course, Google. I’d even venture a guess that in a few quarters, we may be looking at a web where Instagram’s desktop usage is stronger than ChatGPT’s.
The EU/UK analysis shows AliExpress also reaching 20%+ of panelists in those countries.
And in Europe, Reddit’s consistent growth, ChatGPT’s impressive Q4, and Wikipedia’s similar-to-the-US decline are all interesting stories. But let’s move on to the main narrative here: where searches happen across the Internet.
The Headline: Google is responsible for nearly 3/4ths of all desktop web searches; ChatGPT is smaller than you might think
This is the graph I’ve been wanting to see for years — a comparison of the quantity of searches made by millions of devices on dozens of domains, not just traditional search engines and AI tool prompts. And it’s a doozy:
Google alone was responsible for 73.7% of all desktop searches across the 41 domains we analyzed in the US in Q4 2025 (as noted, the graph is not to scale or none of the other label names would be visible). That’s obviously huge, but it’s also far lower than how their market share is usually reported (e.g. Statcounter, whose methodology puts them at 90%+, or our prior, more limited analyses with similar numbers) and higher than what they tried to use in their antitrust defense (i.e. data from Evercore ISI, an “equities research firm”).
Perhaps more fascinating and unexpected are the other domains with more search activity than ChatGPT: Amazon, Bing, and YouTube. Three domains where search marketers historically have put limited effort compared to the onslaught of dollars flooding the “we need to rank in ChatGPT!” space.
If we break all 41 sites into five top-level categories (traditional search, commerce, social, AI, and other verticals), we get an impressively thorough picture of where desktop search behavior actually happens. ~80% of searches still occur on traditional search engines. Commerce sites (Amazon, Walmart, Booking, Airbnb, eBay, etc.) account for ~10%. Social networks come in third at ~5.5%. And AI tools make up 3.2%.
That’s a very different story that what most tech and mainstream media sources are telling us, though, in fairness, Google’s own results are becoming more and more AI-influenced and shaped. That’s mostly through the ~16% of Google SERPs that show AI Overviews, rather than the <0.1% of searchers who click over to AI Mode.
For the EU/UK, the picture is similar:
(please forgive my typo above – this should read “41 Major Sites”; will fix up soon – Rand)
But, in the EU/UK, Google is more dominant (with an additional 5% of all searches vs. the US), while Amazon, YouTube, and Bing hold smaller shares of the search pie. ChatGPT is ~0.75% higher than its US market share, which tracks with Europe’s slightly faster/more aggressive adoption of AI tools overall.
How Share of Search Changed in 2025
The Q4 snapshot, as engaging as it is, doesn’t tell the whole story. We wanted to dig deeper into how search and AI prompting of these platforms changed over the year, which I’ve visualized for the top 7 most-searched properties below:
Unfortunately, Google’s dominance makes reading any other change an eyechart, so I adjusted the Y-Axis to start at 70% of market share rather than 0%, and the trends quickly became clear.
The one I’m most excited about is invisble — it’s the 34 sites outside the top 7 growing their share of search — one of the only areas of web behavior we’ve investigated in the last decade(?!) where the biggest sites aren’t getting more dominant with time. Fingers crossed this trend continues.
Also of note are the growth of Amazon, Bing, and YouTube, while ChatGPT and Google shrank. And not a small amount either — Google lost 3.5 points of share. I can’t think of any other metric by which Google shrank in 2025 (except perhaps consumer perception and employee trust?).
In the EU/UK, the story’s a bit different. The total share of the top 7 is relatively stable. Google shrank by a smaller 2 points of share. And ChatGPT dropped <0.2 points. Amazon’s growth might be the bigger story — EU/UK folks seem to be turning to the e-commerce giant more frequently with their transactional searches.
The Percent of Visitors who Search vs. Browse
A fascinating note (that Amanda teased on LinkedIn last week) that emerged during our research was that many visitors to commonly-searched platforms… don’t search at all. I asked Datos to dig deeper and report on the percent of monthly visitors who performed at least one search or AI prompt on the domain they visited. Here’s the 12-month average of those numbers:
Almost everyone who visits Google searches on Google. DuckDuckGo, Bing, Craigslist, and Amazon are all >70%, but then, it drops off. Only ~half of visitors to eBay and ChatGPT search/prompt at all. Why?
Datos’ panel doesn’t have data about intent (obviously), but their CEO, Eli, noted that many ChatGPT visitors only see chats that others have shared with them; they’re not entering their own prompts. As for eBay, key functions, like category browsing and checking your account’s sales or auctions, don’t require a search, and many who start searching are auto-suggested to visit a category instead via the search bar (which wouldn’t show as a “search” in Datos’ numbers).
It’s another reminder to be wary of using visits as a proxy for usage, especially when it comes to the major AI tools. Visits and prompts are two different things, and if you’re looking at data on traffic-based popularity rather than prompt quantity, you’ll likely be misled.
Speaking of… we at SparkToro have been guilty of this exact fallacy. When I received this data, I immediately emailed my cofounder, Casey Henry, and as of today, we’ve updated our Search and AI Tool usage graphs to be search-focused rather than visit-centric. After all, if you’re investigating an audience, you care about where they search/prompt, not whether they visit and leave (or even just check their API credits).
We will continue to use visitation data for other networks (like social media sites) as usage of/influence from these doesn’t rely on searches. I’ll have a post up soon with more about this product change.
2025 Trend of Searches/Searcher
Last, but far from least, we got to explore how the number of searches each searcher performed (on average) changed over the course of the year. For extra granularity, the Datos team provided me with monthly numbers, so you can see the trend in detail. I broke down the 41 sites into three groups: those with 10+ searches/prompts per searcher per month, those with 5-10, and those with <5.
For the search engines, we’re using searches, and for AI Tools (the only one in the 10+ group being ChatGPT) it’s prompt sessions (not total prompts, as follow-ups are treated as part of the same prompting session, just as multiple search result clicks or visits to dozens of pages of results are treated as a single “search” with search engines). Notably, it’s pretty darn stable. ChatGPT had a slow January, but it’s been steady since then. Google, DuckDuckGo, and Amazon, too. Only Bing saw a real trend – a slight decline as the year progressed.
In the 5-10 searches/prompts per searcher per month group, things get messy. There’s a lot of bouncing around in the numbers, but two standouts are Pinterest (whose growth I wrote about in our State of Search a few weeks ago) and eBay (whose steady decline might be more due to aforementioned autosuggesting of categories and improved browsing UX, but it’s hard to know for sure).
Finally, in the 1-5 searches group, I’m particularly surprised by LinkedIn’s presence here (I expected they’d be a more popular search destination mainly because so many people are looking for, well… people). My guess (and this is just a guess) is that both they and Reddit are low because, while there’s lots of search demand, Google is better at searching LinkedIn and Reddit than their own site search functions. It’s also worth calling out Threads’ remarkable growth, from ~2 searches/searcher to nearly 4 over the course of the year. This is likely a result of heavier use and greater importance of the network: in January 2026, they surpassed Twitter/X’s daily users (at least on mobile).
Important Caveats & Methodology Notes
Whenever I present research, I work hard (as does the Datos team) to be as honest, transparent, and realistic about the results as possible. We don’t have access to every human’s browsing behavior, nor every device, and we (well, Rand) chose these domains editorially, and intentionally excluded things like adult content sites or popular domains where search does take place but didn’t seem relevant (USPS.com or FedEx.com, for example). I’ve tried to compensate by making those decisions and our methodology as clear as possible in this section, and I’m happy to answer questions to the best of my ability in the comments.
Web search vs. AI prompts vs. E-commerce & Social search:
- AI tool prompts can lead to long, complex conversations, many with goals orthogonal to search (programming, image generation, content creation, etc.)
- Searches on e-commerce, social, or content platforms can have lengthy browsing sessions of multiple products after a single query.
- Many social browsing sessions are similar to searches but don’t explicitly use the “search” functionality/URL, and are thus, also excluded from this report
This report treats all searches and prompt sessions equally:
- AI prompt sessions with multiple back-and-forths were counted as a single “search” just as…
- Google searches with dozens of clicks were treated the same as those with zero clicks and…
- E-commerce searches that led to multiple actions vs. one were also treated as a single search
- We did not filter out non-search-relevant AI prompts, navigational searches on traditional engines, or support searches on other web platforms
Desktop only:
- A huge percent of search takes place in mobile browsers and apps, which were not included in this study. We hope to be able to repeat it when these data sources are available to us.
My Conclusions from this Report
It’s important to note that these are my (Rand’s) takeaways from the report. The Datos team likes to be a little more conservative, presenting data over opinions (which is probably very wise in their position as an unbiased deliverer of clickstream information), but I’ve got spicy takes, and a penchant for sharing them, so here goes 😎
- Search happens everywhere. It’s not just Google, Bing, and ChatGPT. Of the 41 sites analyzed in this study, 23 had more than 0.1% market share of search (assuming Google had 5 trillion searches in 2025, like they did in 2024, that’s equivalent to 5B+ searches in 2025).
- Search is a behavior, not a channel. We (at SparkToro) think it’s well past time for SEO to mean Search Everywhere Optimization, and for web marketing professionals to help folks appear in all the places their audience pays attention, not just wherever’s traditional (or overhyped). Seems like a huge missed opportunity to define yourself as only helping brands market in Google and/or ChatGPT.
- Google’s 2025 US market share is closer to ~70% than 90% when platforms beyond traditional search engines are included. It might drop as low as 65% were we to include the long tail of sites in these categories across the mobile web, mobile apps, and the thousands of domains not included in this study. In the EU/UK, Google’s more dominant: ~80% vs. the ~95% in more standard methodologies.
- Most AI Search and AI Answers happen on Google. Even if you combine every prompt on ChatGPT, Claude, Deepseek, and the rest and assume every prompt is a search-equivalent, Google dwarfs them., If 16% of results are showing AI Overviews, Google’s by far the largest AI search tool in both the US and EU/UK by at least an order of magnitude.
- Amazon, Bing, and YouTube still receive more desktop search activity than ChatGPT, despite the latter’s buzz. If you’re worried about AIO/SEO-for-AI/GIEIO, you should be worried about search visibility in those places, too (if their audiences are relevant to your business).
- What counts as a “Search?” remains an open question. AI tool prompts can lead to long, complex conversations, many with goals orthogonal to search (programming, image generation, content creation, etc.) while searches on e-commerce, social, or content platforms can have lengthy browsing sessions of multiple products after a single query. This report treats all these searches equally, but the diversity of intent and experience is important to keep in mind when drawing conclusions.
I’ve also made the PDF of all the images in this report available here.
Special thanks to Belinda Conde, Stanislav Pugach, Teresa Lee, and Eli Goodman from Datos, and to Amanda Natividad for proofreading and adding to this writeup.
The data displayed in this report has been provided by Datos, A Semrush Company. The analysis is based on Datos’s US and EU/UK panels, representing a diverse and statistically significant sample of users, and covers the time frame of (01/01/2025-12/31/2025). For further information please visit Datos’s website and its Privacy Policy.