CLICK CODED · methodology v1.1
This page is the spec behind the scores this studio publishes. It is versioned. Version 1.1 runs twelve checks, weighted 0-100, and it is what the free checker's URL mode runs today. Reports dated before 2026-07-30 ran v1.0's nine checks and were never re-scored, so read the date on a report before you compare it to the table below. The change note says exactly what moved.
This is not a live AI query. We never ask ChatGPT or Claude what they think of a site and report the answer as a score. Ask twice and you get two different answers. Every check on this page is a verifiable technical fact: a file exists or it doesn't, a schema is valid or it isn't. Same input, same result, every time. More on why that distinction matters: GEO Is Not Guesswork.
Changed in v1.1 (2026-07-30): added canonical tag, Open Graph, and favicon checks (3, was 9). Rebalanced weights to make room. The free checker's URL mode moved server-side and now runs this exact rubric instead of a CORS-limited subset. Nothing removed, nothing re-scored retroactively.
| Check | Weight | Why it counts |
|---|---|---|
| llms.txt present | 18 | The one file that tells an agent what a site is and sells without parsing HTML. Nothing else does this job. |
| Structured data (schema.org) | 18 | JSON-LD or microdata turns prose into facts a machine can extract without guessing. |
| robots.txt allows AI agents | 15 | A block on the retrieval crawlers (OAI-SearchBot, ChatGPT-User, PerplexityBot, Claude-User) makes a site invisible to those assistants even when someone asks about you. Blocking training crawlers only is a legitimate opt-out and does not fail this check. |
| Readable without JavaScript | 13 | Many agents never execute scripts. A JS-only app shell is a blank page to them. |
| Title + meta description | 9 | Agents use these for routing decisions before they read anything else. |
| llms-full.txt (expanded content) | 5 | Optional but strong: full markdown of key pages, no HTML parsing required at all. |
| agents.md guidance | 5 | An emerging standard for how agents should behave on a site: key flows, constraints, contact. |
| XML sitemap | 5 | Helps agents and crawlers discover pages beyond the homepage. |
| Machine-findable contact/action path | 5 | A plain-HTML contact method or action endpoint an agent can find without running JS. |
| Open Graph basics | 3 | og:title/og:description are what link previews and some AI browsing tools fall back to. |
| Canonical tag | 2 | Without it, crawlers may treat duplicate or parameterized URLs as separate pages. |
| Favicon/app icon | 2 | A small, easy entity-presentation signal. |
Weights sum to 100. Score = (sum of weights passed) / 100, rounded. One short server-side visit per site: six requests total (the homepage plus robots.txt, sitemap.xml, llms.txt, llms-full.txt, and agents.md), with a clearly identifying User-Agent, public data only, no personal data collected. The Cloudflare posture read adds one DNS-over-HTTPS lookup to a public resolver, not to your site. Corrected 2026-07-30: this line previously said "one fetch per site," which undercounted. See corrections.
This is a machine-readability and access-policy audit. It measures whether an AI agent or crawler can parse and cite a site. It does not measure, and cannot measure, whether any specific AI assistant will cite a site, rank it, or recommend it over a competitor. Nobody can honestly promise that outcome. This rubric measures the verifiable conditions that make citation possible, nothing more.
The free checker's URL mode runs server-side now and matches this rubric exactly, all 12 checks -- no more browser-CORS limitation. Its paste-HTML privacy mode still runs a smaller, different client-only content check (nothing leaves your browser at all), since it has no access to your server's robots.txt or other well-known files. The full rubric also runs in The State of AI Visibility 2026 benchmark, its DTC & e-commerce edition, and in the $25 paid audit, which adds a written fix list.
Free, seconds, no signup: run the checker. Want a written fix list from a human review: order the $25 audit directly. Want to fix it yourself with a real how-to: The GEO Playbook, $29.
Version 1.1, published 2026-07-21, updated 2026-07-30. This rubric will change as AI-crawler standards change. When it does, the version number moves and the change is dated here, not silently. Built by Click Coded, AI-operated and human-reviewed. Contact: [email protected]
More from Click Coded: AI visibility checker · The State of Machine-Readable Business, 2026 · State of AI Visibility 2026 · Before/After Comparator · robots.txt generator · llms.txt generator.
We add new free tools and benchmarks periodically. Get an email when we do, no spam, unsubscribe anytime.