Scoring

Every crawl returns two scores, each out of 100.

Health score

Classic technical SEO health: broken links and status codes, redirects, titles and meta descriptions, canonicals, robots directives, performance, security, and mobile / international signals. This is the number that maps to “will search engines crawl, index, and rank this cleanly?”

GEO score

Generative Engine Optimization — how ready your pages are to be understood and cited by AI search (ChatGPT, Perplexity, Google AI Overviews). It rewards:

• Structured data the engines can parse (Article, FAQ, Product, etc.).
• Semantic HTML that chunks cleanly into answers.
• Answer-readiness — question-style headings and extractable blocks.
• Authorship / E-E-A-T signals (bylines, dates, trust markers).
• Substance — enough real content to be worth citing.

A page that’s great for classic SEO can still score low on GEO, which is exactly why crawlie reports them separately.

Severity levels

Findings are graded error → warning → notice. Use --severity to filter output and --fail-on to gate CI on a threshold. See the CLI reference.

Per-page scores

Beyond the site-wide rollup, crawlie computes per-page Health and GEO scores so you can see exactly which URLs drag the average down — surfaced in the desktop app’s pages table and available in the JSON output.

How the scores are weighted

Each score starts at 100 and is reduced by the findings on the page, weighted by severity. An error costs more than a warning, which costs more than a notice. The site-wide score is the weighted aggregate across all crawled pages, so a single broken page won’t tank an otherwise healthy site — but a systemic issue (a missing title template on every page, say) will show up clearly in both the rollup and the per-page breakdown.

What’s a good score?

As a rule of thumb:

• 90–100 — healthy. Ship it; clean up the notices when convenient.
• 70–89 — solid, with real wins available. Work the prioritized Top Fixes.
• below 70 — something systemic is wrong (missing metadata site-wide, many broken links, or thin/duplicate content). Fix the highest-impact rules first.

GEO scores tend to run lower than Health scores on most sites today, because structured data, llms.txt, and answer-ready content are still uncommon — which is exactly the gap crawlie is built to surface.

Why are Health and GEO separate?

Because a page can be flawless for classic search and still be invisible to AI answer engines. A fast, well-linked page with perfect metadata can lack structured data, an author byline, or answer-ready sections — so it scores high on Health and low on GEO. Splitting the two tells you which kind of work moves the needle, instead of averaging two unrelated problems into one misleading number.