|
|
|
Issue #94 · August 11, 2026
|
This Issue
• llms.txt grew 8.8x to 36,120 sites — but 97% of those files received zero AI requests in May 2026
• Consumer trust in AI search collapsed from 82% to 54% in one year — 27% of brands were misrepresented
• Perplexity cites brands in 13% of responses vs ChatGPT's 0.59% — a 22x gap that changes platform strategy
• Brand web mentions predict AI visibility at 0.664 correlation — backlinks: 0.218. The signal hierarchy has flipped
|
llms.txt Grew 8.8x in One Year. 97% of Files Got Zero AI Requests.
May 2026 · Ahrefs Analysis of 137,000 Domains
An Ahrefs analysis of 137,000 domains found that llms.txt adoption grew 8.8x in twelve months — from 4,088 sites in June 2025 to 36,120 by May 2026. The parallel formats followed similar trajectories: llms-full.txt grew 107x, ai.txt grew 99x. Combined, nearly 39,000 sites have published at least one of these files.
The usage data tells the other half of the story. 97% of llms.txt files received zero AI requests in May 2026. Of the 3% that received any traffic, SEO audit tools accounted for 21.7% of requests — significantly more than actual AI retrieval bots at 1.1%. The major AI platforms send conflicting signals: each publishes their own llms.txt for documentation while directing site owners to rely on robots.txt for crawler management. The Ahrefs analysis characterizes llms.txt as "agent-readiness infrastructure" that currently serves IDE agents and RAG systems — not the general AI search visibility tactic most adoption guides imply.
|
|
Consumer Trust in AI Search Fell From 82% to 54% in One Year — and 27% of Brands Were Misrepresented
2026 · Fractl — AI Search Consumer Trust Study
Fractl's consumer trust study found that confidence in AI search accuracy dropped from 82% to 54% in a single year — a 28-point collapse. The skeptic population grew sixfold, from 3% to 17%. More troubling for brands: 27% of organizations report their brand has been misrepresented in AI-generated responses, yet only 24% have formal AI monitoring processes in place.
The generational split is counterintuitive: Baby Boomers (63%) now trust AI search more than Gen Z (47%), despite Gen Z's lifetime technology exposure. The cohort with the highest familiarity holds AI to the highest accuracy standards. On disclosure: 84-91% of consumers across formats demand AI content labeling, yet only 20% of organizations consistently disclose AI use. For brand teams, the monitoring gap is the most actionable finding — less than a quarter of brands have any system to catch inaccurate AI representations before customers see them.
|
|
Platform Comparison
Perplexity Cites Brands in 13% of Responses. ChatGPT: 0.59%. The 22x Gap That Changes Your Platform Strategy.
A 34,234-response study found that Perplexity cites brands in 13.05% of answers while ChatGPT cites them in just 0.59% — a 22x difference. Perplexity averages 21.9 citations per response vs ChatGPT's 10.4. Grok sits even higher at 27%. Separately, only 11% of cited domains appear on both ChatGPT and Perplexity — confirming these platforms draw from largely separate source pools. A strategy optimized only for ChatGPT is invisible on the other two platforms by default.
Read more →
|
|
|
|
Signal Research
Brand Mentions Predict AI Visibility 3x Better Than Backlinks
Ahrefs' analysis of 75,000 brands found that brand web mentions correlate with AI citation rates at 0.664 — roughly 3x stronger than backlinks at 0.218. The implication cuts against a decade of SEO intuition: link-building campaigns produce the wrong signal for AI visibility. The brands that appear most consistently in AI answers are those with broad mention footprints across Reddit, press, review platforms, and community discussions — not necessarily those with the highest domain authority scores.
Read more →
|
|
|
Navarro's Take
"The llms.txt data is the most instructive story of the quarter. The gap between adoption (8.8x growth) and actual utility (97% zero requests) reveals something fundamental about how GEO signals mature: they get implemented before the infrastructure exists to use them. This is normal — it happened with robots.txt, with sitemap.xml, with structured data. The brands publishing llms.txt today aren't getting citation lift from it now. But if you're waiting until AI crawlers universally support it to implement it, you've already missed the first-mover window. The question is whether you're doing it strategically or just ticking a checklist."
— Ido Navarro, LLMagnet
|
|
|
