Skip to main content
salesolution

AI-Readiness Score · methodology

Every signal, every weight, published.

The AI-Readiness Probe scores a page the way answer engines read it: 42deterministic checks across four categories. This page is the full rubric — generated from the same code that runs the scan, so it can’t drift from what we actually measure. Same URL, same scores, every time.

How the number is computed

  • Each category is a weighted checklist. A signal earns its full points, partial credit, or nothing; the category score is the earned share of the points that apply to that page type, scaled 0–100.
  • The overall score is 0.6 × mean + 0.4 × minimum across the categories we could score. The minimum matters because an answer engine trips on your weakest layer, not your average.
  • We fetch the page HTML (5-second timeout, no JavaScript execution — the same view most AI crawlers get), plus robots.txt and llms.txt. Domain data comes from DataForSEO’s backlink index. Nothing about your page is stored.
  • AI crawlers checked in robots.txt: GPTBot, ClaudeBot, Claude-Web, PerplexityBot, Google-Extended, CCBot. A full block means the engine can’t read you — and a page it can’t read is a page it can’t cite.

Schema

The machine-readable layer. Answer engines parse JSON-LD before they read a word of copy — these checks measure whether yours exists, validates, and carries the properties engines actually use.

  • JSON-LD structured data exists5 pts
  • All JSON-LD blocks parse as valid JSON5 pts
  • A recognized entity type is declared (Product, Article, Organization…)15 pts
  • The primary entity carries its required properties20 pts
  • Citation-grade properties beyond the minimum (brand, ratings, contact, search)15 pts
  • Two or more distinct entity types are described5 pts
  • BreadcrumbList markup places the page in the site hierarchy10 pts
  • Organization/LocalBusiness entity with name, logo, and url10 pts
  • sameAs links point to your profiles elsewhere (LinkedIn, socials)10 pts
  • Open Graph title, description, and image are set5 pts
  • A canonical URL is declared5 pts

AI-readable

Whether a crawler can load real content and follow its structure. This is where JavaScript-only pages, blocked AI bots, and wall-of-text formatting fail. We fetch your robots.txt and llms.txt alongside the page.

  • Page is not blocked by a noindex directive5 pts
  • robots.txt lets AI crawlers (GPTBot, ClaudeBot, Perplexity…) in10 pts
  • An llms.txt file guides AI crawlers (llmstxt.org)8 pts
  • Exactly one H1 states what the page is10 pts
  • Heading levels descend without skips8 pts
  • Content is sectioned with H2/H3 subheadings7 pts
  • Substantive body copy (300–4,000 words)12 pts
  • A <main> or <article> landmark marks the content8 pts
  • Semantic <header> and <footer> frame the page5 pts
  • Meta description in the 70–160 character sweet spot10 pts
  • Title tag of 15–60 characters10 pts
  • At least one heading is phrased as a question AI can answer8 pts
  • Lists or tables give AI extractable structure7 pts
  • Real text is in the HTML, not behind JavaScript8 pts
  • Mobile viewport is declared5 pts

Authority

The trust signals that decide citation. Scored by page type: an article is judged on bylines and dates; a homepage or product page is judged on organization trust — contact, people, reviews — instead. A page is never penalized for signals that don’t belong on it.

  • A named publisher/organization stands behind the page (in schema)12 pts
  • Real-world contact: a phone number AND an email or address12 pts
  • About/contact/team pages are linked in your header or footer10 pts
  • Cites sources: outbound links to 5+ distinct domains12 pts
  • Images carry alt text8 pts
  • Document language is declared4 pts
  • Privacy policy or terms are linked6 pts
  • Linked profiles on other platforms (sameAs or social links)10 pts
  • A named author is creditedarticles only14 pts
  • Published/updated dates are machine-readablearticles only12 pts
  • Shows a machine-readable update datenon-article pages6 pts
  • Named people (Person markup with a role) stand behind the businessnon-article pages10 pts
  • Ratings or reviews are marked upnon-article pages10 pts

Domain strength

Off-page reality, from live backlink data (DataForSEO): domain rank, referring domains, and total backlinks, log-scaled so the first hundred links matter most. Perfect markup on a weak domain still loses the citation — this category is why the score can’t be gamed with markup alone.

  • Domain rank is in the range AI already cites40 pts
  • Other sites link to you (referring domains)35 pts
  • Total backlinks pointing at the domain25 pts

Changelog

  • 2026-07-09 · v1.3 — Domain strength category added (DataForSEO backlink data).
  • 2026-07-09 · v1.2 — Citation-grade bar: llms.txt, question headings, recommended schema properties; weakest-gate overall.
  • 2026-07-09 · v1.1 — Page-type-aware scoring; signal set tripled.
  • 2026-06 · v1.0 — First release: 13 on-page checks.

Citing this rubric

The AI-Readiness Score is Sale Solution’s own rubric, not an industry standard. Reference it as: “AI-Readiness Score, Sale Solution (42-signal rubric), salesolution.net/ai-readiness/methodology/”. The scan is free and needs no email — run it from the homepage.

Score your page against it.

The probe runs this exact rubric in about a second. No email for the scores.

Run the probe →