The click economy is contracting

  • AI Overviews cut clicks to the #1 organic result by ~58% (Ahrefs, 2025), up from 34.5% earlier that year.
  • Zero-click rate ~83% on searches where an AI Overview appears (2026 trackers).
  • AI Overviews appear on roughly 25–60% of searches depending on the tracker, and the share keeps climbing (2026).
  • Generative-AI referral traffic grew several-fold over 2025 — about +700% across the year (≈357% year-over-year by some trackers) — and AI-referred visitors convert above Google organic.
  • Only 38% of AIO citations come from pages ranking in Google's top 10 — down from 76% in prior measurement (Ahrefs, 2026).

Where AI citations actually come from

  • ~94% of AI citations come from non-paid, non-brand-owned sources; a brand's own site accounts for only 5–10% (Muck Rack; McKinsey AI Discovery, 2025–2026).
  • ~28% of pages most cited by ChatGPT have zero Google organic visibility (Ahrefs, 2026).
  • ~80% of sources AI platforms cite don't appear in Google's results; only ~12% match Google's top results (Ahrefs).
  • 67% of ChatGPT's top citations are off-limits to marketers — Wikipedia ~30%, brand homepages ~24%, app stores ~7% (Ahrefs).
  • ChatGPT visibly cites only ~15% of the URLs it retrieves, by some analyses.
  • AI Mode and AI Overviews reach the same conclusions ~86% of the time but share only ~14% of their sources.

Content & format levers (what raises citations)

  • GEO research (Princeton, IIT Delhi, Georgia Tech; KDD 2024): five content moves — citing sources, adding quotations, statistics, fluency, and an authoritative voice — lifted AI citation visibility by roughly 30–40% (exact per-method gains vary by study).
  • A density of 2–3 cited data points per 300 words correlates with higher AI mention rates.
  • Listicles are the most-cited page type in ChatGPT responses at 43.8% (SE Ranking, 2025).
  • Freshness is a strong citation signal — AI engines (Perplexity especially) favor recently updated content.
  • Schema markup has near-zero effect on AI citations — 1,885 pages tracked adding schema across 2025–2026 showed no real uplift (Ahrefs). (Keep it for Google rich results.)
  • The source set behind a Google AI Overview rotates roughly every ~2 days.

Authority & mentions beat links

  • Web mentions outperform backlinks ~3:1 for AI Overview presence — mentions correlate 0.664 vs 0.218 for links (Ahrefs).
  • YouTube presence is the strongest correlate of AI brand visibility across 75,000 brands — ahead of backlinks and domain rating (Ahrefs Brand Radar, 2026).
  • A profile on G2 / Trustpilot / Capterra / Yelp gives ~3× higher odds of being cited by ChatGPT (SE Ranking).
  • A Tier-1 unlinked mention ≈ 60% of a do-follow link's ranking weight (2026 estimate).
  • Third-party mentions and press coverage typically begin influencing AI citations within weeks of publication (estimates range ~2–8 weeks).

Bots & the trust problem

  • Bots = 53% of web traffic; bad bots 40%, the 7th straight year of growth (Imperva, 2026).
  • AI agent traffic +7,851% YoY; AI-enabled bot attacks up ~12.5× (Imperva, 2026).
  • GA4's automatic bot filter only catches the IAB/ABC known-bot list — sophisticated invalid traffic passes through.

The AI search market & infrastructure

  • Google still handles ~80–90% of conventional search; AI engines absorb an estimated 15–20% of informational queries.
  • MCP: ~97M downloads; donated to the Linux Foundation's Agentic AI Foundation, Dec 2025 (Anthropic, with OpenAI & Block) — details.
  • Agentic commerce: Adobe measured +4,700% YoY gen-AI traffic to US retail; the x402 protocol processed ~165M agent transactions in its first months.
  • IndexNow: 80M+ sites, 5B+ daily URL submissions (Bing, Yandex, Naver, Seznam, Yep — Google abstains).
  • Cloudflare: blocks AI crawlers by default for new domains (since July 2025); "Markdown for Agents" cuts ~80% of tokens vs HTML.
  • llms.txt adoption ~6%; major AI crawlers reportedly don't fetch it.
  • WordPress powers ~43% of all websites (~62% of the CMS market).
  • GEO/AEO agency retainers: ~$2k–$8k/mo mid-market, $10k–$25k+ enterprise; AI-visibility tracking tools from ~$29/mo.

What the numbers tell you to do

Three conclusions cut across the data: (1) the click is shrinking, so optimize to be the cited source; (2) ~94% of citations are off your domain, so off-site mentions and entity authority matter more than content volume; (3) the proven on-page levers are quotable, dated, sourced content — not schema markup. For the full method, start with the SEO vs AIO pillar.

Frequently asked questions

How much do AI Overviews reduce clicks?

On queries with an AI Overview, clicks to the #1 organic result fall ~58% (Ahrefs, 2025), up from 34.5% earlier in 2025. The zero-click rate on those searches rises to ~83%.

What share of AI citations come from a brand's own website?

Only ~5–10%. Roughly 94% of AI citations come from non-paid, non-brand-owned sources (Wikipedia, reviews, news, forums). Off-site presence and entity authority matter more than on-site content volume.

Do statistics and citations improve AI visibility?

Yes. Princeton/IIT Delhi/Georgia Tech research (KDD 2024) found five content moves — citing sources, adding quotations, adding statistics, improving fluency, and using an authoritative voice — raised AI citation visibility by roughly 30–40% over baseline (exact per-method gains vary by study). Aim for 2–3 cited data points per 300 words.

Sources, as reported: Ahrefs, SE Ranking, Imperva, Cloudflare, Adobe, Conductor, and the Princeton/Georgia Tech/IIT Delhi GEO paper, 2024–2026. Treat figures as directional trends; re-verify for high-stakes decisions. Last updated June 7, 2026.