The Conversation About Search Has Been Missing Most of the Data
For years, conversations about where people search online have been shaped by a simple, incomplete question: Is it Google, or is it Bing? More recently, that binary has expanded to include a louder, more dramatic version of the same question: Is it Google, or is it ChatGPT?
Both framings miss the point.
In January 2026, SparkToro’s Rand Fishkin, working with clickstream data company Datos (a Semrush company), published a landmark piece of research titled “New Research: Search Happens Everywhere: An Analysis of 41 Websites with Significant Search Activity.” Instead of asking which search engine is winning, the research asks a far more useful question: where do people actually conduct searches across the entire web?
The answer cuts through months of hype, counter-hype, and headline-driven panic. And it has direct, actionable consequences for every marketer, brand manager, content creator, and business owner trying to be found online.
The findings cover a Datos desktop panel of millions of devices across the United States, all 27 EU member countries, and the United Kingdom, tracked across every quarter of 2025. Forty-one domains were analyzed spanning traditional search engines, e-commerce platforms, AI tools, social networks, travel sites, real estate portals, classifieds, and reference sites. No study of this scope and cross-platform breadth has been conducted before.
This post breaks down every major finding, explains what the numbers actually mean for your marketing and content strategy, and answers the questions most frequently asked about search behavior in 2025.
Why This Research Matters More Than Standard Search Market Share Reports
Before diving into the data, it’s worth explaining why standard search market share reports — like those from StatCounter, which puts Google at over 90% globally — tell an incomplete story.
StatCounter and similar services measure market share only among recognized search engines: Google, Bing, Yahoo, DuckDuckGo, Yandex, and a handful of others. They do not count a query typed into Amazon’s search bar. They do not count a product search on eBay. They do not count a recipe search on YouTube, a hashtag lookup on Instagram, a job title search on LinkedIn, or an AI prompt on ChatGPT.
But all of those are searches. A person typing “running shoes under $100” into Amazon’s search field is searching just as meaningfully as someone typing the same phrase into Google. The behavior — query in, results out, decision made — is identical. The platform is different.
This is the gap that SparkToro and Datos set out to close. By analyzing search behavior across 41 domains representing the broad landscape of how people use the internet, the research provides a genuinely complete picture of where search activity happens in 2025.
The Methodology: How the Data Was Collected
The research is built on Datos’ 2025 desktop panel — millions of devices located in the United States and the EU/UK — tracked monthly across the full calendar year. The panel measures actual user behavior, not self-reported surveys or modeled estimates.
Rand Fishkin editorially selected 41 domains from the most-visited 250 domains in Datos’ panel, narrowed to those with meaningful search behavior. The selections were deliberate and transparent: yes to Instagram (where users frequently search for accounts, hashtags, and content), no to USPS.com (where users visit for tracking, not exploratory search). Adult content sites were excluded.
The 41 sites span these categories:
- Traditional search engines: Google, Bing, Yahoo, DuckDuckGo, and others
- E-commerce: Amazon, Walmart, eBay, Etsy, and comparable platforms
- AI tools: ChatGPT, Claude, Gemini, Copilot, DeepSeek
- Social networks: YouTube, Instagram, Facebook, Reddit, Pinterest, LinkedIn, TikTok, Threads
- Reference sites: Wikipedia
- Travel: Booking.com, Airbnb, and travel-focused platforms
- Real estate and classifieds: Zillow, Craigslist, and similar domains
Each domain’s “search” was defined as any query submitted through the platform’s search bar or AI prompt interface. The methodology treats each search or prompt session as a single unit, regardless of how many clicks, follow-up queries, or browsing actions follow.
This is a desktop-only dataset. Mobile browsers and apps — where a substantial portion of all search activity takes place — are not included. The researchers acknowledge this limitation explicitly and note they hope to expand the study when those data sources become available.
That caveat stated, what this study reveals about desktop search behavior is more comprehensive than any comparable research published to date. Source
Google at 73.7%: Still Dominant, But Not at 90%
The headline finding is striking on two fronts.
Google accounts for 73.7% of all desktop searches across the 41 analyzed domains in the United States during Q4 2025. In the EU/UK, that figure rises to approximately 80%.
On one hand, this confirms what everyone already knows: Google is by a significant margin the most-used search destination on the internet. On the other hand, this number is dramatically lower than how Google’s market share is typically reported. StatCounter and similar methodologies put Google at 90% or higher because they only count searches on recognized search engines. When every platform where people search is included, Google’s true share shrinks considerably.
To put this in perspective: if Google processed approximately 5 trillion searches in 2025 — matching their widely cited 2024 figure — then the remaining 26.3% of searches across the other 40 platforms represents over 1.8 trillion additional searches happening somewhere other than Google every year. That is not a rounding error. It is a substantial portion of all human search behavior on desktop.
The EU/UK data also reveals an interesting asymmetry. Google is more dominant in European markets than in the US when measured against all search platforms (approximately 80% vs. 73.7%), but Amazon and YouTube hold smaller shares there than in the US.
Also worth noting: Google lost 3.5 percentage points of search share in the United States over the course of 2025 — a meaningful shift given how rarely any metric has shown Google shrinking. In the EU/UK, the decline was smaller at approximately 2 percentage points. This is not a catastrophic decline, but it is a directional signal that cannot be ignored.
The Full Breakdown: Where Desktop Searches Actually Happen
When SparkToro and Datos categorized all 41 platforms into five groups, the picture became even clearer. Here is where desktop searches actually happen in the US, as of Q4 2025:
Traditional Search Engines: ~80% This category includes Google, Bing, Yahoo, DuckDuckGo, and similar dedicated search platforms. Google alone accounts for 73.7%. Bing holds a meaningful share. Together, traditional search engines still dominate — but they account for 80% of searches, not 90%+.
Commerce Sites: ~10% Amazon, Walmart, eBay, Etsy, Booking.com, Airbnb, and similar transactional platforms together represent approximately 10% of all desktop searches. This is a significant number that rarely appears in search market share discussions. When a consumer searches on Amazon, that behavior has real commercial intent — often more direct buying intent than a Google query.
Social Networks: ~5.5% YouTube, Instagram, Facebook, Reddit, Pinterest, LinkedIn, TikTok, Threads, and similar networks collectively account for 5.5% of all desktop searches. This category is growing. It is also underrepresented in the allocations most marketing teams make.
AI Tools: ~3.2% ChatGPT, Claude, Gemini, Copilot, DeepSeek, and all other AI tools combined represent approximately 3.2% of all desktop searches. This is important to understand clearly: all AI tools together, including the most-discussed ChatGPT, account for just over 3% of desktop search activity.
Other Verticals: ~1.3% This catch-all includes reference sites (Wikipedia), real estate portals (Zillow), classifieds (Craigslist), and other domain-specific search platforms not captured in the above categories.
The 3.2% figure for AI tools is perhaps the most counterintuitive finding in the entire report, given the volume of media coverage those platforms receive. It does not mean AI tools are unimportant — their influence on search behavior and information consumption is real and still growing. But it does mean the conversation about AI “replacing” or “dominating” Google is not supported by data measuring where searches actually happen. Source
Amazon, YouTube, and Bing Each Outpace ChatGPT
One of the most practically significant findings in the research is this: Amazon, Bing, and YouTube all receive more desktop search activity than ChatGPT.
This matters for how marketing budgets and content strategies are allocated. In recent months, a substantial amount of attention — and spending — has moved toward “optimizing for AI” and specifically toward ensuring brand visibility in ChatGPT’s responses. Meanwhile, Amazon, YouTube, and Bing have continued doing what they’ve always done: fielding large volumes of searches with clear commercial and informational intent, from audiences with demonstrated purchasing power.
To be direct: if ChatGPT search visibility is a priority in your strategy and Amazon search visibility is not, the data suggests that priority structure is misaligned with where your customers are actually searching.
This is not an argument against optimizing for AI tools. It is an argument for ensuring that the less-discussed, less-hyped platforms are receiving proportional attention based on actual search volumes. The data shows Amazon SEO, YouTube content strategy, and Bing optimization are all, by the numbers, more impactful channels than ChatGPT optimization — and all three are receiving less marketing attention than the evidence warrants.
How Search Behavior Shifted Throughout 2025
The research tracks not just a single Q4 snapshot but the full trajectory of search behavior across 2025, with monthly granularity provided by Datos. Several trends stand out.
Google’s share declined. From Q1 to Q4 2025, Google lost approximately 3.5 percentage points of US desktop search share. This is notable because it is one of the few metrics showing Google moving backward. The research frames this carefully — it does not necessarily mean people are abandoning Google, but it does mean that search behavior is spreading across more platforms at a measurable rate.
Amazon grew. In both the US and EU/UK markets, Amazon increased its share of searches over the course of the year. This tracks with broader trends in e-commerce search behavior, where consumers increasingly skip Google and go directly to Amazon when shopping intent is clear.
ChatGPT’s growth plateaued. After initial rapid growth, ChatGPT’s share of searches flattened over 2025. The research notes that ChatGPT’s trajectory on desktop positions it likely to end up behind Reddit, Amazon, Facebook, YouTube, and Google in terms of overall reach. In the EU/UK, ChatGPT’s market share is approximately 0.75 percentage points higher than in the US, suggesting slightly faster AI adoption in Europe.
Threads is growing rapidly. Meta’s Threads platform grew from approximately 2 searches per searcher per month in January 2025 to nearly 4 by year-end. This coincides with Threads surpassing Twitter/X in daily mobile users in January 2026, at least on mobile. The growth signals increasing use of Threads as both a discovery and search destination.
Pinterest showed meaningful growth. Pinterest’s searches per searcher grew noticeably over 2025, consistent with separate data in SparkToro’s quarterly State of Search reports. Pinterest occupies an interesting position: its users search with high purchase intent, and its algorithm surfaces content in ways that make organic visibility genuinely valuable for consumer-facing brands.
Wikipedia declined. The reference giant saw declining traffic in 2025, which Rand Fishkin attributes largely to Google sending significantly less traffic to Wikipedia than in previous years. This is a downstream effect of Google surfacing more direct answers via AI Overviews and Knowledge Panels, reducing the need to click through to Wikipedia.
Bing declined. Bing was the only platform in the 10+ searches/searcher group to show a notable downward trend over the year, with a gradual decline in searches per searcher from Q1 to Q4.
The ChatGPT Reality Check: Visits Are Not the Same as Searches
One of the most methodologically valuable insights in the research addresses a measurement error that has been shaping the industry’s understanding of AI tools: most metrics about AI tool usage are based on visits, not actual searches or prompts.
The research reveals that only approximately 50% of people who visit ChatGPT actually submit a prompt. The other half visit but do not search. Many of these visitors are viewing conversations that other users have shared publicly — they consume AI-generated content without ever entering a query themselves.
This single fact changes how every visit-based metric about ChatGPT should be interpreted. If a study says “ChatGPT had X million visits,” roughly half of those visits involved no search behavior at all. Metrics that equate ChatGPT visits with ChatGPT searches are, by this measure, overstating actual prompt volume by approximately 2x.
By contrast, nearly everyone who visits Google performs a search. DuckDuckGo, Bing, Craigslist, and Amazon all see more than 70% of visitors submit at least one search query. The gap between visits and actual search behavior on AI platforms is substantially larger than on traditional search engines.
For marketers and researchers drawing conclusions from visit-based data about AI tool adoption, this is a critical correction. Prompt volume — not visit volume — is the right proxy for actual AI search behavior.
SparkToro updated its own product in response to this finding. As of publication, SparkToro’s “Search and AI Tool” usage visualizations use search-focused rather than visit-centric data to reflect where audiences actually search and prompt, not merely where they land.
AI Search Is Mostly Happening on Google
A point that often gets lost in discussions about AI search is that Google is, by a wide margin, the largest AI search engine in the world.
Approximately 16% of Google search results pages display AI Overviews — AI-generated responses that appear above organic results. Given that Google processes an estimated 5 trillion searches annually, that means roughly 800 billion search result pages per year include an AI-generated overview. No other platform — not ChatGPT, not Perplexity, not Claude — comes close to that volume of AI-assisted responses to search queries.
The narrative that AI search means ChatGPT or Claude ignores the fact that the world’s most frequently used search engine already delivers AI-generated answers at a scale that dwarfs every dedicated AI tool combined.
This matters for content strategy. Optimizing for AI Overviews in Google — through structured content, authoritative sourcing, clear answers to specific questions, and the E-E-A-T signals Google uses to select content for AI-generated responses — is, statistically speaking, the highest-impact form of “AI search optimization” available to most publishers and brands.
That is not to say that visibility in standalone AI tools is unimportant. For certain B2B categories, high-consideration purchases, and research-heavy topics, ChatGPT and Perplexity citations have meaningful influence. But in terms of scale, Google’s AI Overviews touch far more searches. Source
Social Search Is Real, Measurable, and Growing
The 5.5% social search figure in this study deserves more attention than it typically receives.
Social platforms have long been dismissed as not “real” search destinations. The data says otherwise. YouTube is the world’s second-largest video platform and a dominant search destination for how-to content, product reviews, tutorials, and entertainment discovery. Reddit is, by many measures, one of the most trusted sources of peer recommendations on the internet — and its search volume is substantial enough to appear in this research’s top platforms by share.
Instagram’s growth on desktop is one of the more surprising findings. Instagram was historically a mobile-first platform with limited desktop functionality. By 2025, the data shows it has become a genuine desktop search destination — with users searching for accounts, products, hashtags, and branded content. Instagram’s desktop traffic grew consistently throughout 2025, to the point where Rand Fishkin speculates that Instagram’s desktop reach may surpass ChatGPT’s in coming quarters.
LinkedIn occupies an interesting position. Despite being the dominant professional network, its per-searcher search frequency is lower than expected, likely because Google is often a more effective tool for searching LinkedIn’s content than LinkedIn’s own native search — a pattern that also applies to Reddit.
TikTok’s position as a search platform is relevant but less prominent in this desktop dataset given its mobile-first nature. The research’s desktop-only scope means TikTok’s full search impact is likely underrepresented here.
Nearly 40% of Gen Z users, according to separate research cited by Talkwalker, prefer searching TikTok or Instagram over Google when looking for certain types of information. The SparkToro/Datos data, covering all age groups on desktop, still shows 5.5% of all searches on social platforms — a figure that would almost certainly rise substantially in a mobile-inclusive dataset.
E-Commerce Search: What Amazon’s Numbers Mean for Product Marketers
Commerce platforms — led by Amazon, with contributions from Walmart, eBay, Etsy, Booking.com, Airbnb, and others — account for approximately 10% of all desktop searches in the US.
For anyone selling physical products, this number carries direct strategic implications.
When a consumer types a product query into Amazon, they have already made a decision: they are considering a purchase. That is a qualitatively different type of search than the same query entered into Google, which might indicate research, comparison, price-checking, or any number of other intents. The transactional specificity of Amazon search makes it a high-value channel that receives, in most marketing strategy discussions, far less attention than the volume warrants.
In the EU/UK, Amazon’s share of searches grew more than any other platform in absolute terms over 2025. European consumers are increasingly turning to Amazon as a first-stop for product search, bypassing Google for shopping-intent queries.
eBay’s search behavior showed an interesting trend: a gradual decline in searches per searcher, which Datos and Fishkin attribute partly to improved category browsing UX and auto-suggestions that route users away from the search bar and into curated category browsing. This is a reminder that search volume data reflects both user intent and UX design — platforms can influence their own search metrics by making non-search browsing more prominent.
The Percent of Visitors Who Actually Search: A Metric That Changes Everything
The visits-to-searches gap deserves its own section because it fundamentally changes how platform-by-platform comparisons should be made.
Here is the key finding: not everyone who visits a platform searches on it.
The research asked Datos to calculate the percentage of monthly visitors who performed at least one search or AI prompt on each domain. The 12-month average reveals a significant spread:
- Google, DuckDuckGo, Bing, Craigslist, Amazon: All above 70% — the majority of visitors search.
- eBay, ChatGPT: Approximately 50% of visitors search. Half of visitors arrive, browse, or consume content without submitting a query.
- Social platforms and content sites: Generally lower — many visitors browse, scroll, or navigate without using the platform’s search function.
The eBay and ChatGPT data are particularly notable. For eBay, the explanation involves category browsing and account management activity that doesn’t involve a search query. For ChatGPT, the explanation, offered by Datos CEO Eli Goodman, is that many ChatGPT “visitors” are simply viewing shared conversations — they navigate to a URL shared by someone else, read the AI output, and leave without prompting anything themselves.
This is a data quality issue with profound implications. Any analysis that uses ChatGPT visits as a proxy for ChatGPT search behavior is working with a figure that is approximately twice the actual prompt volume. Organizations making strategy decisions based on visit-based AI adoption data may be overestimating the shift toward standalone AI tools.
The correct metric for understanding where people search is not visits — it is search or prompt sessions per platform. On that basis, Google’s lead is even more pronounced than visit-based comparisons suggest, because nearly everyone who visits Google searches on Google. Source
What “Search” Means Across Different Platforms
The research is admirably transparent about a genuine definitional challenge: not all searches are equivalent.
An AI prompt on ChatGPT might be a multi-turn conversation lasting 20 minutes, producing detailed research output, code, creative writing, or complex problem-solving. It might also be a one-sentence question with a two-sentence response. Both are counted as a single “prompt session.”
A Google search might produce a zero-click result — the user reads the featured snippet and leaves. It might produce a browsing session spanning dozens of pages across an hour of research. Both are counted as a single “search.”
An Amazon search might result in a single product purchase. It might result in a 40-product comparison session. Both are counted as a single “search.”
The research counts all of these equally, which is both a limitation and the only methodologically defensible approach when comparing across platforms. The diversity of intent, depth, and duration across search types is important context when drawing conclusions from raw volume comparisons.
What the data shows is where humans initiate search behavior — the act of entering a query into a platform to find something. The downstream value of each search varies enormously by platform and use case. An Amazon search with buying intent may be worth far more commercially than an informational Google search with no buying intent, even though they count equally in this dataset.
Search Everywhere Optimization: What Marketers Should Do With This Data
The core strategic implication of the SparkToro/Datos research can be stated simply: search is a behavior, not a channel.
Rand Fishkin has argued in a companion piece that SEO should now stand for “Search Everywhere Optimization” — a term originally coined by Ashley Liddell of Deviation. The acronym is the same. The scope is different.
Search Engine Optimization historically meant: get higher rankings on Google, Yahoo, Bing, and a few other recognized search engines.
Search Everywhere Optimization means: be visible in all the places where your audience goes to find information, products, services, and answers — which now includes Google, yes, but also Amazon, YouTube, Reddit, Instagram, Pinterest, ChatGPT, LinkedIn, and the dozens of other platforms where meaningful search behavior is documented.
This is not a call to spread marketing resources thin across every platform simultaneously. It is a call to let data drive platform prioritization. The SparkToro/Datos research provides a baseline for that prioritization, at least on desktop.
Here is what the data-driven priority stack looks like for most brands:
Priority 1 — Google (traditional and AI Overview): Still 73.7% of desktop searches. No other channel comes close. Organic SEO, structured data, E-E-A-T, and AI Overview optimization remain the highest-volume search investment for nearly every brand.
Priority 2 — Amazon (for product sellers): 10% of all desktop searches happen on commerce platforms, with Amazon as the dominant player. If you sell physical products and you are not investing in Amazon SEO, you are ignoring roughly 10% of all search behavior — with the highest transactional intent of any category.
Priority 3 — YouTube: YouTube’s search volume exceeds ChatGPT’s. Video content optimized for YouTube search reaches an audience that is not being fully addressed by Google SEO alone. YouTube’s search algorithm favors watch time, engagement, and keyword relevance in titles and descriptions.
Priority 4 — Social Search (Reddit, Instagram, Pinterest, LinkedIn): 5.5% of all desktop searches. Reddit’s search volume is large, and its content frequently surfaces in Google results as well, creating a dual-channel benefit. Instagram and Pinterest have clear search utility for visual product categories. LinkedIn dominates professional search intent.
Priority 5 — AI Tools (ChatGPT, Perplexity, Claude, Gemini): 3.2% of all desktop searches — smaller than the social category. Visibility in AI tools matters, but the scale of investment should reflect the scale of the channel. The most effective way to appear in AI tool responses is still to produce authoritative, well-structured, frequently cited web content — the same foundation that drives Google rankings.
The practical implication is that brands and content teams should audit not just their Google visibility but their presence across each of these categories, calibrated to the 5-category breakdown the research provides.
Key Takeaways from the SparkToro/Datos Research
Here, drawn directly from the original research, are the six conclusions stated by Rand Fishkin:
- Search happens everywhere. Of the 41 sites analyzed, 23 had more than 0.1% market share of search. If Google processes 5 trillion searches annually, that 0.1% threshold equals 5 billion searches per platform — a meaningful number by any definition.
- Search is a behavior, not a channel. The practical definition of SEO needs to expand to include all the places where audiences conduct queries, not just traditional or currently hyped platforms.
- Google’s true US market share is closer to 73-74% than 90% when all major search platforms are included. This could drop toward 65% if mobile app search, the long tail of search-capable sites, and other excluded domains were added.
- Most AI search happens on Google, through its AI Overviews feature. Google is the largest AI search platform by volume, not ChatGPT or any standalone AI tool.
- Amazon, Bing, and YouTube each outpace ChatGPT in desktop search activity. Investment in those platforms is justified by the data in a way that investment in ChatGPT optimization, at current scale, is not — at least as a first priority.
- Defining “search” remains genuinely complex. AI prompts, product searches, informational queries, and social media search all count as search behavior in this framework. Their commercial and informational value varies significantly.
Frequently Asked Questions
What is the SparkToro/Datos “Search Happens Everywhere” research? It is a study published in January 2026 by SparkToro founder Rand Fishkin and Datos (a Semrush company), analyzing search activity across 41 major websites using Datos’ desktop panel of millions of devices in the US and EU/UK. It is the most comprehensive analysis of cross-platform search behavior conducted to date and covers the full calendar year 2025. Source
How much of all desktop search does Google handle? According to this research, Google handled 73.7% of all desktop searches across the 41 analyzed platforms in the US in Q4 2025. In the EU/UK, Google’s share was approximately 80%. This is significantly lower than the 90%+ figure typically cited, because those figures only count recognized search engines and exclude commerce, social, and AI platforms.
Does Amazon really have more search volume than ChatGPT? Yes. Amazon, Bing, and YouTube all receive more desktop search activity than ChatGPT, according to the SparkToro/Datos data. This is one of the most counterintuitive findings in the research and has direct implications for how marketers should prioritize platforms.
Is ChatGPT replacing Google? The data does not support this conclusion, at least through 2025. ChatGPT’s desktop search share is approximately 1-2% in the US as a standalone platform, and its growth rate plateaued over the course of 2025. Google remains dominant. However, Google itself is incorporating AI heavily through AI Overviews, which appear on approximately 16% of Google SERPs.
What percentage of ChatGPT visitors actually use it as a search tool? Approximately 50%. The other half of visitors to ChatGPT are viewing shared conversations that others have published publicly, not submitting their own prompts. This means visit-based metrics for ChatGPT overstate actual search usage by roughly 2x.
What is “Search Everywhere Optimization”? Search Everywhere Optimization (SEO) is a term coined by Ashley Liddell of Deviation and adopted by SparkToro, referring to the practice of optimizing brand and content visibility across all platforms where audiences conduct search behavior — not just Google, but also YouTube, Amazon, Reddit, Instagram, Pinterest, AI tools, and others. The acronym is the same as traditional SEO; the scope is expanded.
What does the research say about social media as a search platform? Social networks collectively account for approximately 5.5% of all desktop searches, according to the research. YouTube, Reddit, Instagram, Pinterest, LinkedIn, and Threads all have measurable search behavior. Threads showed particularly strong growth in searches per searcher over 2025, growing from approximately 2 to nearly 4 per searcher per month.
How did Google’s search share change in 2025? Google lost approximately 3.5 percentage points of US desktop search share over 2025, while Amazon, YouTube, and the 34 smaller platforms outside the top 7 grew their shares. In the EU/UK, Google’s decline was smaller at approximately 2 percentage points. This is one of the first major datasets showing Google moving backward on a search share metric.
Does this research include mobile search? No. The study is desktop-only, using Datos’ desktop device panel. Mobile search — conducted in mobile browsers and apps — is not included. The researchers acknowledge this limitation and note that mobile search would likely change the relative platform rankings, potentially boosting TikTok and Instagram’s social search figures significantly.
How reliable is the Datos data used in this research? Datos is a Semrush company with a large, diverse, and statistically significant panel of desktop devices in the US and EU/UK. The methodology is based on actual observed user behavior (clickstream data), not surveys, modeled estimates, or platform-reported figures. The research acknowledges the limitations of a desktop-only panel and of editorial domain selection, and both Rand Fishkin and the Datos team are transparent about the caveats.
What does this mean for SEO strategy in 2025 and beyond? The research reinforces that SEO strategy should no longer be synonymous with Google-only strategy. Given that approximately 26% of all desktop searches happen outside Google — with 10% on commerce platforms, 5.5% on social networks, and 3.2% on AI tools — brands need to audit and invest in visibility across a broader platform set. The prioritization should be data-driven: Google first (73.7%), then the platforms your specific audience uses most, based on your industry and customer profile.
What platforms are growing their share of search? The research found that 34 out of 41 platforms outside the top 7 are collectively growing their share of desktop search. Amazon and YouTube showed clear growth within the top 7. Threads showed the most proportional growth among smaller platforms. Pinterest’s searches per searcher also grew noticeably. This is described in the research as one of the few areas of web behavior where the biggest platforms are not becoming more dominant over time.
How does EU/UK search behavior differ from US search behavior? In EU/UK markets, Google is more dominant (~80% vs. 73.7%), while Amazon holds a smaller slice of search than in the US. ChatGPT’s share is approximately 0.75 percentage points higher in EU/UK, suggesting slightly faster AI tool adoption in Europe. AliExpress also reaches 20%+ of EU/UK panel devices, reflecting its popularity in European markets.
What is the correct way to measure AI tool adoption for search? According to this research, the correct metric is prompt sessions per platform, not visits. Because approximately half of ChatGPT visitors do not submit prompts, visit-based comparisons overstate ChatGPT’s search utility. The research prompted SparkToro to update its own product to display search-focused metrics rather than visit-centric ones.
Where can I read the original SparkToro/Datos research? The original study is published in full at SparkToro.com. A PDF of all the charts and graphs from the report is also available via the research post.
What This Research Changes — And What It Confirms
The SparkToro/Datos study does not tell marketers to abandon Google. It tells them something more nuanced: that Google’s dominance is real but narrower than most reporting suggests, that the “Google vs. ChatGPT” framing misses most of where search behavior actually occurs, and that the data-driven path forward involves platform prioritization based on where audiences genuinely search — not where the current news cycle says they should be.
The finding that Amazon, YouTube, and Bing each outpace ChatGPT in desktop search volume will not generate the same headlines as “ChatGPT threatens Google.” But it is more accurate, better sourced, and more directly useful to anyone making decisions about content strategy, SEO investment, or digital marketing allocation.
The finding that Google processes most AI-assisted searches — through AI Overviews affecting 16% of SERPs — is similarly underreported. If your goal is to show up in AI-generated answers, Google is still the highest-priority target, not standalone AI tools.
The finding that 34 platforms outside the top 7 are collectively growing their search share signals something worth watching: the search landscape is becoming more distributed over time, even if slowly. Brands that build visibility only on the most obvious platforms risk being invisible precisely where their audience searches for them.
None of this is speculative. It comes from a desktop panel of millions of devices, tracked monthly over a full year, across 41 carefully selected platforms — the most comprehensive study of its kind ever published.
This post draws on original research by Rand Fishkin and Datos (a Semrush Company), published as “New Research: Search Happens Everywhere; an Analysis of 41 Websites with Significant Search Activity” at SparkToro, January 2026. All statistics cited are sourced directly from that research.
About ALM Corp
ALM Corp is a digital growth and marketing consultancy that helps businesses build measurable visibility across every platform where their customers search, discover, and decide. The SparkToro/Datos research on cross-platform search behavior reflects exactly the kind of evidence-based marketing strategy ALM Corp is built around: instead of chasing headlines about which platform is winning, ALM Corp uses data to identify where your specific audience actually searches — whether that is Google, Amazon, YouTube, Reddit, Instagram, or the growing range of AI tools — and builds content, SEO, and distribution strategies calibrated to that reality.
In a landscape where 26% of all desktop searches happen outside Google, where Amazon outpaces ChatGPT in search volume, and where social platforms collectively represent 5.5% of all queries, “Search Everywhere Optimization” is not a trend to watch. It is the operational framework ALM Corp applies to every client engagement. If your current strategy treats Google as your only search channel, we can help you measure what you are missing and build a roadmap to address it — with data, not hype.
Learn more at www.almcorp.com
