What this check tests
The audit takes the list of pages it crawled and checks whether every page's URL begins with https://. If all do, the check passes with a score of 100. If any does not, it fails with 0. There is no partial.
The URL it tests is the address the audit asked for, not the address the response finally came from. The crawler follows redirects, but it records the requested URL. The other crawled pages are normally picked from your sitemap and homepage links on the same origin, so they share the scheme of the address the audit started with. In practice the result mostly comes down to whether the address given to the audit started with https://. There is one exception. If fewer than three pages were crawled, the audit adds up to five more from your sitemap, matching the host name only. If the sitemap lists http:// URLs for those, they are fetched, their requested http:// addresses are recorded, and the check fails even for an https:// audit. The platform adds https:// to an address typed without a scheme, so a fail usually means someone entered an http:// address on purpose.
This has consequences worth knowing.
- A site that serves everything over HTTPS passes. A site whose
http://address redirects to HTTPS still fails if you audit thehttp://address, because the redirect is followed but not recorded. - The check does not look at the certificate, the TLS version, mixed content (an HTTPS page that loads HTTP resources), cookie flags or whether plain HTTP is redirected at all. If the HTTPS handshake fails, the audit cannot fetch the homepage, and you get an unreachable-site result instead of a score for this check.
- The Security headers check is where HSTS is tested.
Why it matters to agents
HTTPS is the baseline for any client that sends requests or reads responses over a network it does not control. Google's web.dev guidance says HTTPS helps prevent intruders from tampering with communication between your site and its users, and that intruders exploit every unprotected resource, including images, cookies, scripts and HTML. It also says HTTPS is required for many new browser features.
For agents the practical points are narrower. An agent that fetches a plain http:// page cannot tell whether what it received was changed in transit. Addresses in your sitemap and links that use http:// also send crawlers through an extra redirect. This page cites no vendor document that says an AI assistant ranks or cites HTTPS sites more often, so treat the weight 2 as a security and trust floor, not a visibility lever.
How to fix it
Most hosting platforms issue certificates and serve HTTPS automatically. If yours does not:
- Get a certificate for every hostname you serve, including
wwwand any subdomain, and install it. web.dev walks through the key, the certificate signing request and the installation. - Enable HTTPS on the server and confirm it works before changing anything else. web.dev warns against enabling HTTPS only to redirect clients back to HTTP.
- Redirect all
http://requests to thehttps://version of the same URL. web.dev says that you should ultimately redirect HTTP to HTTPS and use HSTS. - Update internal links, your sitemap, canonical tags and
og:urlvalues to usehttps://. See Canonical URLs. - Fix mixed content: load images, scripts and styles over
https://. - Once everything works, add a
Strict-Transport-Securityheader. RFC 6797 defines it as a policy that tells user agents to turn insecure references to your host into secure ones and to end connections that have certificate errors. Start with a shortmax-ageand raise it as you gain confidence.
To verify, run curl -sI http://example.com/ and look for a 301 or 308 with a Location that starts with https://. Then run curl -sI https://example.com/ to confirm the 200. If a page on your site returns errors after the move, Status codes will list them.