Backed by HTTP Archive Web Almanac 2024 — and hundreds of Rankmio audits. The honest state of the web, without sugar-coating.
Data study on 16.9 M websites (HTTP Archive Web Almanac 2024) combined with 729 Rankmio audits (Jan–Jul 2026). The core result: seven of ten websites fail AI visibility, 99 % lack FAQ schema, only 41 % of the web uses JSON-LD. Full methodology below.
Your customer types a question into ChatGPT. Four competitors get cited. Your website doesn’t appear. Not because you’re worse. But because your site speaks the wrong language — the language of machines.
The analysis of Rankmio audits shows: this is not the exception, it is the norm. Combined with the neutral reference of the industry — the HTTP Archive Web Almanac 2024 covering 16.9 million websites — a clear picture emerges.
What follows is an inventory of five structural weaknesses that affect the majority of the web. Each finding is backed by two independent data sources.
Two independent data sources combined: a broad public dataset (Web Almanac) and a targeted proprietary audit set (Rankmio). Overlap validates each other.
Rankmio audits: January to July 2026. Web Almanac reference: June 2024 crawl, published November 2024.
Rankmio: 74 external domains, 85 free-check reports, 729 individual pages with detailed GEO page scoring. Web Almanac: 16.9 million websites, crawled desktop and mobile.
Mixed B2B and B2C: online commerce, professional services, medical practices, mid-sized manufacturing, agency client projects. Predominantly German-speaking market.
Web Almanac SEO and Structured Data chapters (public, methodology documented). Rankmio Free SEO Check aggregate and GEO Page Score analysis (proprietary, page-level signals).
Composite score (0-100) combining structured data completeness, entity linking, author attribution, freshness signals, and topical coverage. Threshold for «visible»: ≥ 70.
Sub-score (0-100) measuring the probability of being cited as source by generative AI. Weights source authority, quotability, factual density, and machine-readable structure.
The Rankmio sample is not randomly selected — it reflects websites whose operators actively sought an audit and therefore tend to have SEO awareness. Real-world numbers on unaudited sites are likely worse. Web Almanac data is global and may vary by region. Both sources are snapshots and do not track individual sites over time.
The Rankmio audit sample and the Web Almanac public dataset converge on the same five structural weaknesses. Where both sources measure the same signal, deviations are small and confirm each other.
no JSON-LD structured data
In Rankmio audits, Schema.org markup is missing on every fourth customer. Google and AI systems literally don’t know what the page is about.
Source: Web Almanac 2024 (16.9 M websites)no FAQ schema
FAQPage adoption is at 0.6 % globally. On product pages in Rankmio audits: 100 % without FAQ schema. On legal pages: 83 %.
Source: Web Almanac 2024 + Rankmioimages without alt text
Almanac median: 58 % of images have alt text — so 42 % don’t. In our audits, 49 % of pages show broken alt attributes. Congruent.
Source: Web Almanac 2024 + Rankmiono canonical tag
Almanac: 31 %. Rankmio: 24 %. That means Google guesses which URL is canonical — and often guesses wrong.
Source: Web Almanac 2024 + Rankmiono single H1 tag — or several
Almanac: 30 %. Rankmio: 20 %. Google and LLMs read the H1 as „what is this actually about“. Without it, the machine decides on its own.
Source: Web Almanac 2024 + RankmioThese numbers aren’t cosmetics. They are the reason many websites carry good content and remain invisible in AI-generated answers. The data can be illustrated with a simple analogy:
Imagine a library where the books have no spines. The content is excellent, but no one can figure out what’s on which page. That is a website without Schema.org, without clean headings, without structural signals.
For Google, this is inefficient. For ChatGPT and Perplexity, this is the end of the journey.
Not all page types are equally affected. We broke down 729 audited pages by category. The results are counter-intuitive.
The irony: the pages meant to sell (products) are worst documented. The pages that should signal trust (legal) practically don’t exist for Perplexity and Gemini. Blog content — where you might expect the biggest gaps — is the strongest area. The care that flows into editorial content is apparently higher than for transactional pages. If you’re wondering where to start: not with the blog. With the money pages.
Classic SEO mistakes are uncomfortable. But the part that really hurts is new.
The Web Almanac confirms this indirectly: only 1.77 % of pages use Article schema, only 1.5 % Product schema. When AI systems scan the web for citable sources, they find nothing to hold onto in 98 % of cases. If you’re among those, the AI picks other sources — and you’re missing from the answer.
That’s the real 2026 problem: Google ranking still matters. But it’s increasingly a side discipline.
The audit data indicates three levers with the highest impact-to-effort ratio. They address the structural gaps identified above.
Mechanical work, but the effect is huge. Rankmio detects which pages have schema gaps and generates the JSON-LD markup directly from existing content.
Content Studio →20 % of pages have fewer than two internal links, 11 % are complete orphans. Rankmio’s link graph shows where the bridges are missing.
Link Graph →The citation tracker weekly checks whether your domain gets cited by ChatGPT, Gemini, Perplexity and Claude. Only measurement enables improvement.
Citation Tracker →The advantage doesn’t sit with the giants — it sits with those who realized first that search has shifted. You can see that shift as a threat. Or as an open window. Because as long as the competition still ignores the numbers, you have time. Just not forever.
Run your own reality check →Free SEO & GEO Check
SEO score, AI visibility and citability of your website in 30 seconds — no registration required.
Register for free, get 10 credits and start right away.
Register now