How the scoring works

What each score actually measures.

Every score is built from concrete checks we run on your pages — HTTP status, canonical and noindex tags, rendered vs. raw content, internal links, structured data, and index reality. No black box: each number ties back to specific findings you can fix.

What each score means
0–100

Retrievability Score

One headline number for whether AI and search can reach, render, index, and cite the site.

  • Blends four sub-scores: Reach (can bots fetch it), Render (do they get content, not an empty JS shell), Index (is it really getting indexed at the corpus level), Cite (is it citation-ready).
  • A high score does not mean guaranteed ranking, citation, or traffic.
  • Use it as triage: what is ready, what needs review, and what is invisible.
0–100

Indexability Score

Measures whether search engines can crawl, parse, and consider the page for indexing.

  • HTTP status, fetchability, canonical tags, noindex, visible content, metadata, internal links, and JavaScript risk.
  • Blockers such as noindex or non-200 responses heavily reduce the score.
  • Clean pages become candidates for sitemap/Search Console/IndexNow/Bing workflows.
0–100

GEO Score

Measures whether AI answer engines can identify, summarize, and cite the page.

  • Entity clarity, answer-style content, source/methodology signals, structured data, trust pages, and visible context.
  • A good GEO page clearly explains what it is, who it helps, what claims it makes, and where its data comes from.
  • The goal is citation-worthiness, not keyword stuffing.
0–100

Agent Score

Measures whether AI agents and crawlers can understand the site structure and content.

  • Semantic HTML, structured data, visible initial HTML, metadata, crawler policy, and llms.txt readiness.
  • Agent-ready does not mean every bot will use the site.
  • It means the site exposes clean signals that make machine reading easier.
How pages are classified

Blocker

A page should not be submitted aggressively until fixed. Examples: noindex, fetch failure, non-200 status, robots blocking.

Warning

A page can sometimes be submitted, but the issue may reduce indexing odds. Examples: thin content, weak internal links, canonical ambiguity.

Info

Useful improvement, but usually not a hard indexing problem. Examples: weak description or missing optional enhancement.

The standards behind every check

We don't invent ranking factors.

Every signal maps to published guidance from Google, Bing, Schema.org, or an open protocol — linked so you can verify it.

Index & crawl checks

  • Returns HTTP 200. Google only indexes pages that return a successful 200 response. Errors, redirects, soft-404s and auth walls are dropped from the index. Google · HTTP status codes & Search →
  • Page is fetchable. If Googlebot can't fetch the URL (DNS, SSL, firewall or bot-blocking), it can't be crawled or indexed at all. Google · Crawling overview →
  • No accidental noindex. A `noindex` robots directive tells Google to keep the page out of search entirely — a common accidental blocker on pages you actually want indexed. Google · Block indexing with noindex →
  • Has a canonical tag. A self-referencing canonical helps Google consolidate signals to the right URL and avoid treating variants as duplicates. Google · Consolidate duplicate URLs →
  • Canonical points to itself. When a page's canonical points elsewhere, Google may index that other URL instead of this one. Intentional sometimes, but often an error. Google · Consolidate duplicate URLs →
  • Clear, unique title. The title is one of the strongest relevance and click signals, and Google uses it to generate the title link in results. Google · Influence your title links →
  • Useful meta description. Google often uses the meta description for the result snippet; a clear one improves click-through even though it isn't a ranking factor. Google · Control your snippets →
  • Enough substantive content. Thin or near-duplicate pages are the most common reason Google crawls a URL but declines to index it. Helpful, original content is the fix. Google · Helpful, people-first content →
  • Connected by internal links. Google discovers and ranks pages partly through internal links. Orphaned or weakly-linked pages get crawled less and indexed slower. Google · Make your links crawlable →
  • Content visible without heavy JS. If important text and links only appear after heavy client-side JavaScript, Google may index a near-empty page. Server- or static-render the key content. Google · JavaScript SEO basics →

GEO & AI-agent factors

  • Structured data. Schema.org markup (JSON-LD) that lets search engines and AI assistants understand the entities on the page rather than guessing from text. Google · Intro to structured data →
  • Semantic HTML. Use of main/article/section/nav/header/footer so crawlers and AI agents can parse the page's structure reliably. MDN · HTML semantic elements →
  • Answerable content. Clear questions and definitions, cited sources or statistics, structured data, and last-updated signals — the kind of content generative engines quote. We weight cited evidence over raw length on purpose. This is an emerging best practice, not a guaranteed ranking factor. Google · Helpful, people-first content →
  • llms.txt / AI crawler policy. An llms.txt file and explicit AI-crawler rules in robots.txt tell AI systems what they may use. A proposed standard adopted by a growing number of sites. llmstxt.org · The /llms.txt proposal →

GEO is a fast-moving area. These are best-practice signals, not guaranteed citation factors — because no engine publishes one.

Important

Scoring is diagnostic, not predictive.

A clean page is eligible for discovery and indexing workflows. Search engines and AI systems still decide what to crawl, index, rank, quote, or ignore. Our job is to remove preventable technical issues and show the next best action.