Antileak's health score is a weighted 0-100 roll-up of the measured checks from one scan: what an AI crawler receives from your raw HTML, whether the crawlers are allowed in, structured data field by field, the fundamentals, and email authentication on the domain. The governing rule is that only measurements count — a check that could not run contributes nothing, rather than a guessed zero that would read as a finding.
The inputs are the scan's six groups of checks: readable text and business facts in the raw HTML; which of the six named AI crawlers robots.txt blocks; the seven structured-data fields, counted as populated or not; fundamentals — HTTPS, mobile viewport, title, meta description, H1, response time, page weight, broken links among sampled internal links; presentation — canonical, Open Graph, social image, sitemap, image alt text; and SPF, DKIM and DMARC on the domain. Security headers are graded separately as a 0-100 score with A-F bands, so one number never hides the other.
A scan that never reached the homepage reports that it measured nothing, instead of returning zeros. A certificate expiry the platform cannot read is reported as unknown, not estimated. This is why a report can look shorter than a competitor's: nothing is padded with guesses, so the score moves only when a measurement does — which is also what makes two scans of the same site comparable over time.
Across the 2,528 small-business sites in our published corpus the average health score is 53, and 33% of sites score under 50 — the range where sites measurably lose visitors and visibility. So a score in the 50s is normal and unimpressive; the score's job is not to flatter but to tell you which measured building blocks are missing, each of which is individually fixable. Your own number takes about a minute on the free scan.
No. The health score is the 0-100 roll-up of all measured checks; the security score is a separate 0-100 with letter grades at fixed bands — 90 an A, below 60 an F — built from the five response headers and the transport. Keeping them separate keeps both readable.
Because a measurement changed — the score has no other input. Diff the two reports' checks; the moved measurement is the explanation, not a recalibration on our side.
Above the corpus average of 53 puts you ahead of the typical small-business site; under 50 is the range where sites measurably lose visitors and visibility. More useful than the number is the list of which specific checks failed.
Yes, in one narrow way: the score measures what a crawler can retrieve from your site, not whether any assistant's index contains you or chooses you. It measures the part you control.
Where does your site stand? The scan is free, takes about 60 seconds, and needs no account. It fetches your homepage exactly the way GPTBot, ClaudeBot and PerplexityBot do and tells you what they receive.