MeasureBoard.com
All research
AI SearchPublished May 8, 2026

State of AI Search 2026

How AI search engines (ChatGPT, Perplexity, Gemini, Claude) are sending traffic to websites in 2026, based on data from MeasureBoard-tracked properties.

Headline finding
[X.X]%
of all measured website sessions in Q1 2026 came from an AI search engine
Calculated as (sessions from AI sources) / (total sessions) across MeasureBoard-tracked properties for the period below.

Dataset

Sample size
[N] websites tracked by MeasureBoard
Time period
January 1 – March 31, 2026
Coverage
Worldwide

Aggregated, anonymized session-level data from MeasureBoard-tracked GA4 properties. No personally identifiable information is included.

Key findings

[XX]%
of AI-sourced sessions came from ChatGPT
Largest single AI source. Perplexity and Gemini follow.
[X.X]%
median share of total traffic from AI sources across MeasureBoard-tracked sites
Half of all tracked sites are above this number, half below. Weighted aggregates can be skewed by very high-traffic outliers.
[+/-XX]%
QoQ change in AI-sourced sessions from Q4 2025 to Q1 2026
Measured against the same set of properties present in both quarters.
[XX.X]%
of brand-related queries return a citation to the brand's own website on at least one major AI platform
Measured across queries tracked in MeasureBoard's AI Rank Tracker.
[XX]/100
median GEO Readiness Score for sites tracked by MeasureBoard
GEO Readiness combines technical foundation, content structure, schema presence, AI visibility, and content depth.
[X.X]%
of sites have published an llms.txt file
llms.txt is a proposed standard for declaring AI-crawlable content; adoption is still emerging.

Methodology

Sessions are bucketed by source/medium pairs reported by GA4. We classify a session as AI-sourced when the referrer matches a known AI platform (ChatGPT, Perplexity, Gemini, Claude.ai, Copilot, You.com, Phind, and others maintained in MeasureBoard's AI_PLATFORMS registry).

Properties with fewer than 100 sessions in the reporting window are excluded to reduce noise. Bot traffic flagged by GA4's standard filters is excluded.

All percentages are weighted by total session volume — i.e. a site with 1M sessions counts proportionally more than a site with 1k sessions. Median (per-site) figures are reported separately where indicated.

Why this matters

AI search engines route traffic differently than Google. A user who asks ChatGPT “what's the best small-business analytics tool” sees a synthesized answer with 3-5 cited sources — not a ranked list of ten blue links. The economics of being one of those cited sources, versus being on page two of Google, are not comparable.

Most analytics tools group AI-sourced traffic into “Other Referrals” or miss it entirely. The numbers above come from a sample where AI sources are explicitly tracked.

How to act on this

  • Audit your AI search visibility. Run brand-related queries through ChatGPT, Perplexity, and Gemini and check whether your site is cited. MeasureBoard's AI Rank Tracker automates this.
  • Check your GEO Readiness Score. The technical signals AI crawlers look for are different from traditional SEO — schema markup density, structured content hierarchy, llms.txt presence.
  • Publish original data. AI engines disproportionately cite primary sources. A blog post saying “ChatGPT is growing” is interchangeable; a stats page with a number journalists can quote is not.

How to cite this research

Quoting from this page or any of the statistics above? Please link back to this page using the citation below.

MeasureBoard, "State of AI Search 2026," May 2026, https://www.measureboard.com/research/state-of-ai-search-2026

Licensed under CC BY 4.0 — free to use with attribution.