AI Search Traffic Report 2026: 16× Growth & Market Shifts

AI search traffic grew 16× from 2024 to 2026. What was once a rounding error — 0.02% of all website traffic in 2024 — is now 0.32%, meaning roughly 1 in every 312 website visits originates from an AI search engine. By mid-2026, AI referral traffic represents approximately 1.08% of all global web traffic, growing at ~1% per month. This report breaks down the data from SE Ranking's study of 101,574 websites across 250 countries over 16 months, supplemented by the latest Axis Intelligence and StatCounter data through April 2026.

The AI search market in 2026 looks fundamentally different from 2025. ChatGPT remains dominant but is losing share. Gemini overtook Perplexity. Claude is surging in both general and B2B segments. Google's grip on search has loosened for the first time in two decades. And the May 7, 2026 ChatGPT link update dramatically reshaped the citation referral landscape — ChatGPT referral traffic surged 157.7% week-over-week, with homepage referrals up 354.7%. For brands investing in GEO, these shifts directly affect which platforms to prioritize.

July 2026 snapshot: the latest signals

Since the April–June 2026 datasets above were compiled, three fresh signals confirm the trend is accelerating rather than plateauing:

The practical read: the 16× growth from 2024 to 2026 is not a one-time step change but a persistent monthly climb. Brands that refresh content on a 30-day cadence capture the 3.2× citation multiplier (ConvertMate/Semrush 2026) while the overall pool keeps expanding — compounding visibility gains against a rising tide. A query that earns a Perplexity or ChatGPT citation today sits in front of a measurably larger audience than the same citation did six months ago.

August–September 2026 update: the pool keeps expanding, but the click does not

Fresh mid-2026 signals confirm AI search is now a mass behavior, not a niche. Google AI Mode surpassed 1 billion monthly active users by mid-2026 (Google I/O 2026, Gemini 3.5 Flash), and Gemini reached ~900M MAU on the same trajectory. Independent GEO research (Omnius AI Search & GEO Report 2026) puts total AI search at ~45B monthly sessions worldwide, with 80%+ of Google searches ending without a click. The implication for the 16× growth curve is direct: the absolute audience behind every AI citation is materially larger than three months ago, so a fixed citation share now delivers far more impressions — and content freshness matters even more.

But 2026 also delivered the counter-evidence, and it is the most important thing in this report. The largest longitudinal study of AI Overview economics to date — Seer Interactive (April 2026), tracking 2.43 billion impressions across 53 brands and 5.47 million queries over 14 months — found organic CTR on AI Overview queries fell from 1.76% to 0.61% between June 2024 and September 2025, a 65% collapse, before rebounding to 2.4% by February 2026. The rebound is real but the structural gap is not: on the same query, pages with no AI Overview earn 33,500 clicks per million impressions, cited brands earn 20,743, and uncited brands earn just 9,445 — under a third of the pre-AIO baseline.

"Being cited is not a vanity metric. On an identical AI Overview query, cited brands earned 120% more organic clicks per impression than uncited brands — the citation itself is the traffic asset, not a consolation prize for the click you lost."

Meanwhile the zero-click ceiling has hardened. SparkToro's 2026 Similarweb clickstream study found 68% of US Google searches now end without any click — up from roughly 45% a decade ago — rising to ~83% when an AI Overview is present and 93% in Google AI Mode. Treat this as the structural context for every number below: AI search is delivering more impressions and more revenue, while delivering proportionally fewer clicks.

September 2026: Google starts keeping the click it used to send

The most consequential 2026 development for traffic reporting is not a growth number — it is a product change. On 21–22 September 2026, Search Engine Roundtable documented Google testing anchor links inside AI Overviews that open AI Mode instead of the publisher's page. The control still looks like a link. The destination is Google's own conversational surface, not your URL. Citation chips and "Show more" links were already a leaky click path; this test makes the leak deliberate for follow-up queries.

That UI test lands next to the strongest experimental evidence yet assembled. A pre-registered field experiment by Wang, Gleason, Bart, Wilson and Metaxa (UPenn / Northeastern, 2026) tracked 1,100 participants across seven treatment days and found that forcing searches into AI Mode cut external click-through by 18.8 percentage points versus current Search. The authors declare no competing interests and the work is NSF-funded. Pew Research Center's independent measurement points the same direction: visitors click a traditional result on 8% of visits to search pages carrying an AI summary, versus 15% without one — while clicks on links inside the summary sit at 1% of visits.

How to read this together: the 16× growth curve in referral traffic is real and still compounding, but the per-impression economics of a citation are deteriorating by design. Two operational consequences follow. First, report AI search in impressions and branded-search lift, not sessions — the session count is now partly a Google product decision rather than a demand signal. Second, audit claim fidelity separately from citation placement: 11.0% of 98,020 atomic AI Overview claims were unsupported by the pages the summary cited, which means being cited is not proof your content was paraphrased faithfully. Where your number did appear in the answer and the answer was wrong, that is a reputation event worth a correction request, not a traffic win.

October 2026: traffic that never runs a search at all

Six in seven ChatGPT answers never look anything up. Ahrefs analysed 66.6 million free, logged-out ChatGPT responses from August and September 2026 and found that a web search ran in only 14.4% of them (Ahrefs, 12 October 2026). When the engine did search, it averaged just 1.6 fan-out queries. This is the most important number in AI traffic reporting this quarter, because it splits every referral projection in two: the 16× curve measures only the one answer in seven that went to the web, while the other six were written from what the model already knows.

The same study measured where retrieval actually looks. 6.8% of fan-out queries used the site: operator, and among the most frequent of those targeted searches 75.9% pointed at a brand's own website — apple.com alone drew 9,442 site: queries, ikea.com 7,361 and nike.com 5,210. The vocabulary moved the same way: "official" appeared in 13.1% of English fan-out queries in September, up from 10.1% in August, and "price" rose from 6.04% to 8.88%. In a separate 20,000-prompt US test, 84.4% of factual questions cited at least one official source against 68.3% for comparison questions, and 63.4% of all factual citations pointed at official sites, falling to 46.4% for comparisons.

The citation mix moved with the queries, not with the traffic. Comparing ChatGPT's 200 most-cited US domains in March and September 2026, Ahrefs found PubMed rising from 64th to 4th, the FDA entering at 8th, and WebMD falling from 18th to 79th — while Reddit's share of cited answers collapsed from 19.1% to 2.3% and Wikipedia's from 14.1% to 2.9%. Brand support pages, regulators and hands-on testers gained; explainers and aggregators lost. For a topical resource site the lesson is blunt: if a page only restates what a primary source already publishes, the engine increasingly cites the primary source instead.

15 October 2026: Google moves the link, and licenses the news

Google's 15 October 2026 update is the first product change this year that pushes traffic toward publishers rather than away from them. Inline links inside AI Mode and AI Overviews now appear more frequently and higher in the answer; a test places the one or two official pages that complete a specific task in a dedicated block at the top of the response, each with a note explaining why it is useful; the end-of-answer link block gains image thumbnails; and timely-topic carousels expand to more countries and languages while surfacing a reader's Preferred Sources, usage of which Google says has grown 7× since January 2026.

The same announcement expanded Google's News AI pilot past 400 publications, adding CNN, the Financial Times, Axios and Forbes alongside local US newsrooms. Licensing and traffic are separate questions: publishers are paid for content delivered to the experimental features, not by how much their material contributes to an answer. And the measurement gap survives — Search Console's generative AI performance report still shows no click data months after launch, so a change in link prominence is, for now, unverifiable from the publisher side.

Mentions are not citations: the 13 October measurement reset

Semrush's 2026 AI Visibility Index (13 October 2026; seven monthly measurements from January to July, 2.5M UK and 25M US prompts per month across ChatGPT, Gemini, AI Mode and AI Overviews) put a number on the gap this report has tracked all year: fewer than half of the top 100 mentioned brands are also among the top 100 cited domains, in every month and in both markets.

The most useful illustration is a pair of direct competitors. Tesco and Asda both stayed in ChatGPT's top 10 mentioned UK brands in every month studied — yet in July Tesco was the 25th most-cited UK domain and Asda was 606th. Being named and being sourced are separate outcomes with separate failure modes, and no single aggregate "AI visibility score" reveals which one you have. The report also finds news sources take 29.5% of ChatGPT's top-100 UK citations against 19.6% in the US, and that AI-referred visitors convert 60% better while earning 53% more revenue per visitor (Adobe data cited in the report).

The authenticity tax: what the traffic race costs content

A 14 October 2026 survey of 500+ US and UK marketing leaders (The Typeface Signal Report: The AI Visibility Gap) found 62% say brands are sacrificing content quality to earn AI citations and 61% say the same about brand authenticity — rising to 72% among teams producing the most machine-facing content. 68% are already creating content experiences for AI that humans will likely never see, and 49% plan to deprioritize or stop investing in traditional SEO.

Read that carefully rather than triumphalistically. It is a self-assessment by marketers, not a measurement of page quality, and it is vendor-sponsored research with no disclosed sample size or margin of error. But it names the failure mode this report's data predicts: if six in seven answers never search, and are written from what the model already believes about a category, then a market where every brand publishes the same optimised paragraph does not produce more citations — it produces a category with nothing left to choose between. The defensible position is content that is first-hand — your own tests, prices and dates — not content that is merely maximally extractable.

Methodology: 101,574 websites, 250 countries, 16 months

The SE Ranking AI Traffic Research Study (2026) is the largest publicly available dataset on AI search referral traffic. It tracked 101,574 websites across 250 countries and territories over a 16-month period (January 2025 – April 2026), with 2024 data as a historical baseline. The study monitored referral traffic from 10 AI platforms: ChatGPT, Gemini, Perplexity, Microsoft Copilot, Claude, DeepSeek, Grok, Poe, Mistral AI, and other AI sources.

Traffic share is measured as the percentage of total website visits attributed to each AI platform. Year-over-year growth compares 2025 full-year averages against 2026 averages (January–April). Only platforms exceeding 0.10% of AI traffic in 2026 are included in the detailed analysis: ChatGPT, Gemini, Perplexity, Copilot, and Claude. Additional data from Axis Intelligence (StatCounter, Similarweb, Ahrefs compilation, June 2026) provides the most current market snapshot.

AI search traffic growth: 2024 → 2026

Source: SE Ranking AI Traffic Research Study (2026). 2026 data covers January–April. Axis Intelligence (2026) includes expanded measurement window capturing broader AI search activity. Peak month was October 2025 at 0.3511% of all web traffic per SE Ranking.

The single largest month-over-month jump occurred in September 2025, when AI referral traffic surged 51% in one month. ChatGPT drove approximately 90% of that growth. The second-largest shift came from the May 7, 2026 ChatGPT link update, which increased ChatGPT referral traffic 157.7% week-over-week and homepage referrals 354.7% — confirming that platform-level UX changes, not just user growth, drive citation traffic.

Market share by AI search engine (2026)

Source: Axis Intelligence AI Search Statistics 2026 (StatCounter April 2026 data; growth rates from Similarweb Sep 2024–Mar 2026).

"The AI search market is fragmenting rapidly while simultaneously improving in conversion quality. ChatGPT remains dominant but declining; Gemini is the fastest-growing platform (+9×); Claude delivers the highest conversion rate (16.8%) and overperforms dramatically in B2B (18.5% referral share vs. 2.66% globally)."

B2B AI search: Claude's hidden strength

A critical dimension that aggregate market share data obscures is B2B-specific performance. According to the Goodie Wave 2 study (March–April 2026 average), the B2B AI referral landscape differs dramatically:

Source: Goodie Wave 2 (Mar–Apr 2026 average), compiled by Axis Intelligence. Overperformance ratio = B2B share / global share.

Claude's B2B overperformance (7× its global share) is the most striking data point. If your business serves B2B audiences, Claude may be delivering significantly more referral value than its 2.66% global share suggests.

ChatGPT: dominant but declining in share

ChatGPT remains the overwhelming leader with 76.85% of AI referral traffic in April 2026 (StatCounter), down from 84.21% a year earlier. Its absolute traffic grew +84% year-over-year (Similarweb Sep 2024–Mar 2026). The September 2025 spike (51% MoM) was the single most consequential moment in the dataset, followed by the May 2026 link update.

ChatGPT's 2026 average traffic share is 0.2384% of all website traffic (SE Ranking), up modestly from 0.2286% in January 2026 to 0.2321% in April 2026. The platform has matured from explosive growth to steady expansion. However, the May 7, 2026 link update dramatically changed the game: ChatGPT referral traffic jumped 157.7% week-over-week, with homepage referrals up 354.7%, and 28.3% of ChatGPT-cited pages had zero organic visibility — meaning ChatGPT Search is discovering content that traditional search engines miss entirely.

Gemini: the new #2

Gemini's traffic grew +9× from September 2024 to March 2026 (Similarweb), the fastest absolute growth multiplier of any AI search engine. Its global share reached 9.00% (up from 2.31% in April 2025). In January 2026, Gemini overtook Perplexity globally as the #2 AI search engine by referral traffic.

Gemini showed a notable counter-trend in December 2025: while other AI engines saw traffic decline that month, Gemini grew +52% month-over-month. This resilience suggests Gemini's Google ecosystem integration provides a different — and possibly more durable — growth engine than standalone AI search apps. Gemini now processes approximately 500M+ weekly queries with ~900M monthly active users (Google I/O 2026).

Claude: the fastest-growing engine

Claude's referral traffic surged +770% year-over-year (Similarweb Sep 2024–Mar 2026) with MAU growing +640% — the highest growth rate of any AI search engine. While its absolute share (2.66%) is still small, the trajectory is steep. Claude's B2B share (18.5%) is a remarkable 7× its global share, making it disproportionately important for B2B publishers.

"Claude's 770% growth rate and 18.5% B2B referral share make it the most underappreciated AI search platform in current GEO discourse. Its 200K context window and deep research capabilities make it uniquely suited for citation from long-form, authoritative content — particularly in technical and professional domains."

Perplexity: citation-density leader despite traffic shift

Perplexity's global share declined from 12.07% (Apr 2025) to 7.73% (Apr 2026), and its US AI traffic share dropped from 19.73% to 6.85% during the same period. However, its absolute user base grew to 100M+ MAU — the share decline reflects overall market expansion of 36%, not user loss.

Despite the share decline, Perplexity remains the citation-density leader with 5–15 numbered references per answer — more than any other AI engine — and references placed at the top of each answer rather than below. Estimated source CTR is 18–22%, significantly higher than pre-May-2026 ChatGPT. For brands prioritizing citation volume and brand exposure over raw referral traffic, Perplexity remains the strongest platform.

In 2026, Perplexity launched the Comet browser (iOS, Android, Mac, Windows), Deep Research (state-of-the-art on Google DeepMind Deep Search QA), and Perplexity Computer (19+ model orchestration) — all creating additional citation surfaces for optimized content.

Google AI Overviews: 48% coverage and growing

Google AI Overviews now appear in 48% of all tracked Google queries as of February 2026, up from 31% a year earlier — a +58% year-over-year increase. Coverage varies dramatically by query type:

One methodological caveat matters before you quote any single figure. Every major tracker measures a different query universe, so the headline percentages legitimately disagree — and they are all correct for their own sample. The defensible planning range is 20–50%:

Sources: Semrush 10M+ keyword panel; Conductor Q1 2026 benchmark (21.9M queries); BrightEdge Generative Parser (Feb 2026); Google official statement (Feb 2026); SparkToro/Similarweb via Search Engine Land (Jun 2026). Divergence reflects query-sample composition, not contradiction.

Source: Axis Intelligence AI Search Statistics 2026. Google AI Overview coverage data from February 2026.

If the trend continues, AI Overview coverage is projected to exceed 60% of all queries by early 2027. However, Google AI Mode (the full AI-native experience) has a 93% zero-click rate, meaning cited sources receive dramatically less referral traffic despite being referenced. Independent measurement confirms the gap: Pew Research (68,879 tracked searches from 900 US adults) found only 8% of visits with an AI Overview clicked a traditional result (vs. 15% without one), and just 1% clicked a source inside the AI Overview — so AIO's value is brand visibility and citation presence, not clicks. The strategic implication: being cited in AI Overviews is valuable for brand visibility, but the referral traffic payoff is minimal compared to standalone AI search engines like ChatGPT and Perplexity.

The upside is that AI Overviews cite a far wider slice of the web than organic results ever did. BrightEdge's February 2026 data found only 17% of AI Overview citations come from pages ranking in the organic top 10 — down sharply from 76% in mid-2024. Ahrefs' citation analysis breaks the distribution down further: roughly 38% from positions 1–10, 31% from positions 11–100, and 31% from beyond position 100. AirOps' 2026 State of AI Search report puts ~60% of citations outside the organic top 20. Practically, a page buried on page four can still shape what Google tells your entire category — and a page-one ranking no longer guarantees you are in the answer at all.

Engagement: AI sessions are significantly longer and higher-converting

AI search users spend more time on destination websites and convert at dramatically higher rates than organic search users. The Digital Bloom study (February 2026) quantified the gap:

Source: Digital Bloom conversion referral study (Feb 2026); Similarweb ChatGPT referral panel (Apr–May 2026).

The conversion rate comparison is even more striking:

• ▸ Claude — 16.8% conversion rate (Digital Bloom, Feb 2026) — 9.5× Google organic baseline
• ▸ ChatGPT — 14.2–15.9% conversion rate
• ▸ Perplexity — 10.5% conversion rate
• ▸ Gemini — 3.0% conversion rate
• ▸ Google organic baseline — ~1.76–2.8%

Source: Digital Bloom AI search conversion study (Feb 2026). AI search platforms convert at 1.7–9.5× the rate of Google organic search.

Regional breakdown

AI search adoption varies significantly by region. The US leads in AI traffic share, followed by the EU and UK:

Source: SE Ranking AI Traffic Research Study (2026). EU AI traffic declined 12% within 2026 (Feb→Apr), suggesting possible market saturation or seasonal variation.

Google vs. AI: the narrowing gap

Google still sends 134× more visitors than all AI platforms combined (as of 2026), down from ~300× in early 2025. Organic search still accounts for 42.75% of all website traffic in 2026. But the trajectory is clear: Google's multiplier advantage halved in one year.

Google's search market share dropped below 90% for the first time in 2026 (StatCounter). Google referral traffic to publishers declined −33% globally and −38% in the U.S. in 2025. Meanwhile, 35% of U.S. consumers now start product discovery with AI (vs. 13.6% for traditional search engines). Gartner's 2024 projection of a 25% decline in traditional search volume by 2026 now appears conservative.

The disappeared: DeepSeek and Grok

Not all AI search engines sustained their growth. DeepSeek peaked at 0.37% share of AI traffic in early 2025 but dropped to 0.01% by April 2026. Grok showed brief spikes but returned to near-zero. Neither platform is a meaningful referral traffic source as of mid-2026.

What this means for GEO strategy

The 2026 data changes the GEO playbook in four ways. Underlying all of them is a structural reality measured by the Princeton GEO study (KDD 2024): adding named statistics lifts AI visibility by ~33% and external citations by ~28% — so the traffic gains below are amplified when content is also optimized for citation, not just discovery.

1. 1.
   Optimize for more than ChatGPT

   ChatGPT is still 76.85% of AI traffic, but its share is declining. Gemini (+9×), Claude (+770%), and the growing AI Overviews (48% coverage) represent a combined 23%+ of the citation surface. GEO strategies that only target ChatGPT miss a quarter of AI referral traffic — and a growing percentage.

2. 2.
   Invest in long-form, B2B content for Claude

   Claude's 19-minute average session, 16.8% conversion rate, and 18.5% B2B share make it the highest-value engine for professional audiences. Claude's 200K context window allows it to process and cite from long documents. In-depth research reports and comprehensive guides are ideal for Claude citation.

3. 3.
   Track regional and B2B differences

   AI traffic share varies from 0.29% (US) to 0.16% (UK). ChatGPT's dominance ranges from 71.33% (US) to 85.28% (UK). Claude's B2B share is 7× its global average. If your audience is regional or B2B-focused, prioritize the engines that lead in those segments.

4. 4.
   Refresh content frequently

   Content updated within 30 days receives a 3.2× citation multiplier (ConvertMate/Semrush 2026). This single signal outperforms all Princeton GEO strategies combined in terms of measurable impact. Every article on this site is updated with fresh data — and you should do the same.

5. 5.
   Never let one engine be load-bearing

   ChatGPT holds 76.85% of AI referral traffic, which makes single-platform dependence a live business risk rather than a theoretical one. In August 2026, Reddit's share of ChatGPT Search citations fell 86.4% in four days (3.83% → 0.52%, Promptwatch) after a query fan-out change — while Google AI Overviews declined only 11.3% over the same window. Reddit did nothing wrong. Spread citation equity across owned content, third-party coverage, and multiple engines, and read our full analysis in AI Citation Volatility 2026.

6. 6.
   Publish the fact, not just the explanation

   With 84.4% of factual questions citing at least one official source and 75.9% of ChatGPT's most frequent site: searches targeting brand domains (Ahrefs, 12 October 2026), the highest-yield page type has changed. Your own numbers — tests, benchmarks, pricing, dated observations — now beat a well-structured summary of somebody else's. A restatement competes with the original; the original is what the engine is learning to prefer.

Frequently asked questions