Google’s real market share isn’t 90%. It’s closer to 70%. That gap — 20 percentage points of search activity — represents billions of queries happening somewhere else entirely.
Google’s real market share isn’t 90%. It’s closer to 70%. That gap — 20 percentage points of search activity — represents billions of queries happening somewhere else entirely.
That’s the central finding from a year-long joint research project between SparkToro and Datos (a Semrush company), published in March 2026 by Rand Fishkin. The study analyzed 41 websites with significant search activity, covering US and EU/UK desktop panels across all of 2025. And if you’re a demand generation leader who has spent the last two years debating ChatGPT’s threat to Google traffic, the data has a different message than you might expect.
## The 90% Figure Was Always Wrong
Traditional market share methodologies — tools like Statcounter — measure search volume only across recognized search engines. That’s a reasonable starting point, but it misses an enormous slice of how people actually find things online.
When Fishkin and the Datos team broadened the definition to include all search behavior across 41 major digital properties, Google’s Q4 2025 US market share came in at 73.7%. Significant, yes. But not 90%. And if you extend the analysis to the long tail of mobile apps, mobile web, and thousands of smaller sites not captured in the study, Google’s actual share could drop as low as 65%.
In the EU/UK, Google is more dominant — roughly 78–80% — but still well below the 95% figure that conventional methodologies produce.
This isn’t a minor statistical discrepancy. For a CMO building a demand generation strategy, the difference between 90% and 70% market share changes how you allocate budget, where you build content, and which platforms you’ve been ignoring.
## Where the Other 30% Lives
The research breaks down search activity across the 41 analyzed sites into four main categories for the US market in Q4 2025:
– **Traditional search engines:** approximately 80% of all searches (led by Google at 73.7%, with Bing accounting for much of the remainder)
– **Commerce sites** (Amazon, Walmart, eBay, Booking, Airbnb, and others): roughly 10%
– **Social networks:** approximately 5.5%
– **AI tools:** 3.2%
That commerce figure deserves a longer look. Amazon, in this dataset, has more search activity than Bing. More than YouTube. Certainly more than ChatGPT. If your product sells through or competes with anything on Amazon’s platform, and you don’t have an Amazon search optimization strategy, you’re ceding ground to competitors who do.
Of the 41 sites analyzed, 23 captured more than 0.1% of total search market share. Assuming Google processed 5 trillion searches in 2025 — consistent with 2024 figures — 0.1% market share equals 5 billion searches. Twenty-three platforms, each with over 5 billion equivalent searches annually. That’s not a footnote. That’s a parallel search ecosystem most marketing organizations are barely touching.
## The ChatGPT Attention Problem
Here’s where the data gets uncomfortable for anyone who has spent significant budget or organizational energy on “ranking in ChatGPT.”
In the US desktop dataset, ChatGPT receives less search activity than Amazon, Bing, and YouTube — all three. Despite the volume of coverage, conference sessions, and strategic pivots dedicated to AI search tools over the past two years, the actual search volume flowing through ChatGPT is smaller than commonly perceived.
ChatGPT does show up in an interesting way: it’s the only AI tool with more than 10 prompt sessions per searcher per month, which means its users are highly engaged when they do use it. But roughly 50% of monthly visitors to ChatGPT don’t enter a prompt at all — they’re viewing shared conversations, not actively querying. Visits and searches are not the same metric, and conflating them has led to significant misallocation of marketing attention.
To be direct about it: Bing and YouTube each likely warrant more dedicated search optimization effort than most B2B marketing teams currently give them. That’s a straightforward resource allocation question, and the data makes a strong case for rebalancing.
## Google Is Winning the AI Search War — Quietly
The narrative that AI tools are displacing Google hasn’t materialized in the 2025 data. Quite the opposite.
Approximately 16% of Google SERPs now show AI Overviews. Fewer than 0.1% of searchers click into Google’s AI Mode. When you add up all prompts on ChatGPT, Claude, Deepseek, and every other AI tool, Google still dwarfs the combined total — by at least an order of magnitude in both the US and EU/UK markets.
Google isn’t losing to AI search. Google is becoming AI search.
That said, AI Overviews are creating a real and measurable problem for content publishers. Organic click-through rates for informational content are dropping by up to 64% in zero-click environments, and outbound clicks from Google results are being cut roughly in half. The traffic is still going to Google — it’s just not leaving Google. For demand gen teams that built their inbound engine on Google organic traffic, this is a structural problem that won’t self-correct.
The appropriate response isn’t to abandon Google optimization. Google remains dominant for commercial and local searches — maps, reviews, “near me” queries, transactional intent. The better response is a dual strategy: traditional SEO alongside Generative Engine Optimization (GEO), which focuses on appearing within AI-generated answers rather than just in the blue links below them.
## Search Is a Behavior, Not a Channel
The most durable insight from this research is definitional, and it has real implications for how demand generation teams structure their work.
Fishkin’s core argument: search is a behavior that happens everywhere, not a channel owned by Google or any single platform. When someone types a query into Reddit’s search bar, that’s search. When a buyer searches for software reviews on G2 or Capterra, that’s search. When a LinkedIn user searches for a person or a topic, that’s search — even if LinkedIn’s native search is, by the data’s suggestion, inferior to simply Googling “ site:linkedin.com.”
This reframing matters for B2B demand generation specifically. High-intent users who go directly to a site’s search bar are 5x more likely to convert than users arriving through broad discovery channels. On-site search, enterprise search, and platform-specific search optimization are not UX features — they’re conversion infrastructure.
The implication for demand gen strategy is that “Search Everywhere Optimization” — appearing in all places where your buyers are actively querying — is a more complete framework than SEO alone. It requires asking: where does my target buyer search when they’re in research mode? Where do they search when they’re ready to buy? Those answers probably include YouTube, Amazon, LinkedIn, Reddit, and several industry-specific platforms that never appear in a standard SEO audit.
## What the Year-Over-Year Trends Actually Show
A few data points from the 2025 annual comparison are worth flagging for anyone building a 2026 strategy.
Google lost 3.5 percentage points of search market share in the US across 2025. In the EU/UK, the decline was smaller — approximately 2 percentage points. These are meaningful shifts, even if Google remains dominant.
Of the 34 smaller sites in the dataset (outside the top 7), collective search share grew. The largest platforms are not becoming more dominant — which is unusual in web behavior trends and suggests a genuine fragmentation of search activity across the ecosystem.
Wikipedia showed significant traffic decline throughout 2025, attributed largely to reduced Google referral traffic — a direct consequence of AI Overviews absorbing queries that previously sent users to external sources. That’s a preview of what reduced Google referral traffic looks like at scale, and it’s not theoretical.
Instagram’s desktop traffic grew substantially across the year. Within a few quarters, Instagram’s desktop usage may exceed ChatGPT’s — a comparison that would have seemed implausible two years ago but reflects where actual user behavior is heading.
## The Honest Caveats
This research is desktop-only. Mobile represents a massive portion of search activity — likely the majority for many categories — and is not captured here. The findings are directionally significant but not a complete picture.
The 41 domains were editorially selected by Fishkin from the most visited 250 domains in the Datos panel. This is a reasonable methodology for a study of this scope, but it means the long tail of smaller, industry-specific platforms isn’t represented. For B2B demand generation, some of the most important search behavior may be happening on platforms that didn’t make this particular list.
And the definition of “search” in this study treats all queries equally — a zero-click Google search for a quick fact and a 45-minute Amazon product research session are counted the same way. Intent matters, and intent isn’t captured here.
These aren’t reasons to dismiss the findings. They’re reasons to treat them as a starting point for rethinking search strategy rather than a definitive benchmark.
## Rebalancing Where Attention Goes
For demand generation leaders managing significant budgets and cross-functional teams, the practical takeaway from this research is about resource allocation and strategic honesty.
The marketing industry has spent two years in an intense conversation about AI search — specifically about ChatGPT — while Amazon, YouTube, and Bing have been quietly processing more search volume than any AI tool on the market. The conversation has been disproportionate to the data.
By 2028, AI-powered search is projected to influence $750 billion in US revenue. That’s a real and significant shift coming. But the platforms that will carry most of that volume — Google with AI Overviews, Amazon with AI-assisted product discovery, YouTube with AI-surfaced video — are not the platforms getting the most strategic attention right now.
The brands that adapt well won’t be the ones that pivoted hardest toward ChatGPT optimization in 2024. They’ll be the ones that mapped where their buyers actually search — across all 41+ relevant platforms — and built presence accordingly. The data for that mapping now exists. The question is whether marketing organizations will use it to challenge their own assumptions, or continue optimizing for the search landscape they imagined rather than the one that actually exists.
