Most "AI search statistics" articles circulating in 2026 are unsourced - recycled percentages with no underlying study, presented with false precision. This roundup takes the opposite approach. Every number below carries a named source, a date, and a methodology note. Where credible studies disagree, we cite both rather than averaging them into fake certainty.
The state of AI search in 2026 is genuinely consequential for SEO strategy, but the data is messier than the marketing literature suggests. Here is what we can actually defend.
Adoption: How Many People Actually Use AI Search?
The honest answer is that nobody has a definitive number, and the credible estimates vary by a factor of three depending on what "use" means.
ChatGPT weekly active users. OpenAI confirmed in December 2024 that ChatGPT had 300 million weekly active users. Sam Altman publicly cited approximately 800 million weekly active users at OpenAI DevDay in October 2025, with the figure reported around 900 million by early 2026. This is a self-reported metric, but it is the most widely cited adoption signal.
Google AI Overviews coverage. Google announced at I/O 2024 that AI Overviews would expand broadly across U.S. queries, with subsequent expansion to additional countries. Independent tracking from BrightEdge reported AI Overviews appearing on roughly 48% of monitored queries through 2025 and into 2026, with significantly higher prevalence in informational verticals such as healthcare and education (often above 80% coverage).
Perplexity scale. Perplexity reported approximately 230 million monthly queries in August 2024, growing to over 780 million monthly queries by May 2025. The company does not consistently break out weekly active users.
What is clear: AI search is no longer a fringe behavior. What is not yet clear, and where you should be skeptical of any specific claim, is the precise share of total search volume now routed through AI assistants versus traditional search.
Click-Through Rate Impact: The Disputed Number
This is where the most consequential SEO debate lives, and where the data legitimately disagrees.
| Study | Finding | Methodology |
|---|---|---|
| Ahrefs (March 2025) | Roughly 34.5% lower CTR on the top organic result when an AI Overview is present | Compared CTR for the same queries with and without AI Overviews |
| Semrush (August 2024 / updated 2025) | AI Overviews present on approximately 13% of queries in initial study; CTR impact varied by intent | Tracked SERP composition across a large query set |
| Pew Research (July 2025) | Users clicked on a search result on 8% of visits where an AI Overview appeared, versus 15% on visits without | Behavioral study of actual user sessions |
These are not the same finding. Ahrefs measured CTR on a per-query basis. Pew measured behavior at the session level. Semrush measured prevalence and impact across intents. The honest summary: AI Overviews reduce click-through rates measurably, but the precise magnitude depends on how you measure.
What does not appear to be in dispute: informational queries - particularly definitional, comparative, and how-to intent - show the largest CTR declines. Transactional and navigational queries are far less affected.
Referral Traffic: Where AI Search Sends Users
Traffic referred from AI assistants is small in absolute terms but growing rapidly.
Similarweb's 2025 referral analysis showed that traffic referred from ChatGPT to external sites grew substantially through 2024 into 2025, but remained well below 1% of overall web referral traffic. Perplexity referrals grew on a similar trajectory, off a smaller base. Google AI Overviews referrals are particularly difficult to measure because Google AI Overview clicks are attributed to organic, not to a distinct AI Overviews source in most analytics tools.
The practical implication for SEO: if you are measuring AI search impact by referral traffic alone, you are looking at the wrong metric. The bigger effect is the traffic that does not arrive - the queries that resolve entirely inside the AI Overview without the user clicking anything.
Brand Citation Behavior Across Platforms
Different AI assistants surface brands differently, and the patterns matter for visibility strategy.
Profound and BrightEdge (and several other tracking tools that emerged in 2024 and 2025) have published reports describing meaningful divergence in which sources different AI assistants prefer. Recurring themes in the published research:
- •ChatGPT historically over-indexes on Reddit, Wikipedia, and established editorial publishers - a function of training data and citation sourcing.
- •Perplexity publishes citations in-line for nearly every response, making it the most transparent platform to measure on.
- •Google AI Mode and AI Overviews lean on Google's own index, which favors content with strong traditional SEO fundamentals.
- •Gemini is Google's product on Google's infrastructure; Copilot is Microsoft's product built on OpenAI's GPT models running on Microsoft Azure and Bing. The two run on entirely separate stacks, and their citation behavior differs accordingly by query type.
This is not a single ranking system. Optimization is platform-aware now, the same way mobile SEO became platform-aware in the late 2010s.
Enterprise Adoption of AI Search Strategy
Adoption of dedicated AI search optimization inside marketing teams is real but uneven.
Gartner's 2024 strategic predictions projected that organic search traffic would decline meaningfully by 2026 as users shifted toward AI-mediated answers. The exact percentage Gartner cited has been widely repeated and sometimes misattributed; the underlying point - that AI assistants are pulling material query volume away from traditional search - is consistent with the CTR studies above.
Bain's 2024 GenAI survey reported that a significant majority of large enterprises had piloted generative AI initiatives, with marketing among the most common use cases. AI search visibility as a discrete budget line is newer; most enterprises are still folding it into existing SEO or content budgets rather than creating dedicated teams.
What the Data Does Not Yet Tell Us
A few important questions remain genuinely unanswered in the public research as of mid-2026:
- •Per-platform conversion rates. We know AI search refers traffic. We do not yet have reliable cross-industry data on whether that traffic converts at higher, lower, or comparable rates to traditional organic.
- •Citation half-life. Once an AI assistant cites a source, how long does that citation persist? Anecdotal evidence suggests it varies considerably by platform; published research is thin.
- •Local intent behavior. Most published AI search studies focus on informational and commercial queries. Local intent (the historic Google strength) and how AI assistants handle "near me" queries is under-studied.
If you encounter a statistic on any of the above presented with high precision, treat it skeptically until the underlying methodology is published.
How to Use This Data
Three takeaways for SEO and content strategy:
1. AI Overviews are real and they reduce click-through, but the magnitude is contested. Build content that earns the citation inside the Overview rather than relying on the post-Overview click. 2. Platform-specific optimization is real. ChatGPT, Perplexity, Google AI Mode, and Gemini each have distinct citation behavior. Tracking visibility on one is not a proxy for the others. 3. Direct referral traffic is the wrong primary KPI. Measure citation share, brand mention share, and AI Overview presence on target queries - not just sessions referred from ai.com or perplexity.ai.
Methodology & Sources
- •Ahrefs (2025): CTR study on AI Overview presence, March 2025 publication.
- •Semrush (2024-2025): AI Overviews prevalence and intent study, original publication August 2024 with subsequent updates.
- •Pew Research Center (July 2025): AI Overviews user behavior study.
- •BrightEdge: Generative Parser AI Overviews tracking, ongoing 2024-2025.
- •Similarweb (2025): Cross-platform AI referral traffic analysis.
- •Profound: AI citation tracking research, 2024-2025.
- •OpenAI public statements (2024): Weekly active user disclosures.
- •Perplexity public statements (2024-2025): Monthly query volume disclosures.
- •Gartner (2024): Strategic predictions on organic search decline.
- •Bain & Company (2024): Enterprise GenAI adoption survey.
If a number does not appear in the article, it is because we could not source it confidently. That gap is the point.