The 68 checks, and what each one means
One page per group. Every check is listed with what it watches, what a change in it costs you, how many consecutive runs confirm it before you are emailed, and whether it starts switched on.
Indexability checks 12 checks
The twelve checks that answer one question: can a search engine still index this page? HTTP status, redirects, robots directives, canonicals, sitemap membership, and what Googlebot specifically receives.
On-page tag checks 11 checks
Eleven checks on the tags that decide how a page is presented rather than whether it is indexed: title, meta description, H1, hreflang, structured data, Open Graph and the rest.
Tracking tag checks 2 checks
Two checks on the tracking IDs you declare: whether each one is still in the page source, and whether any now appears more than once. One mechanism, every vendor.
Site configuration checks 9 checks
Nine host-level checks: the HTTPS and canonical-host redirects, soft 404 detection, sitemap and robots.txt reachability, llms.txt, favicon and declared key assets.
Domain and DNS checks 10 checks
Ten checks on the records underneath the site: domain expiry, registrar, nameservers, the email trio of MX, SPF and DMARC, plus DKIM, CAA and DNSSEC.
TLS certificate checks 5 checks
Five checks on the certificate: days to expiry, the issuing authority, whether the chain is complete, whether it still covers your hostname, and which TLS versions the server accepts.
Content and integrity checks 9 checks
Nine checks on the body of the page: outbound link count and page language for hack detection, mixed content, security headers, and four threshold checks you point at something specific.
What Google reports 6 checks
Six checks read from Google Search Console once a day for every watched page: whether Google has it indexed, which canonical Google chose, how its last fetch went, and when it last crawled.
Google's numbers 4 checks
Four checks on Google Search Console's own clicks and impressions: a site or a watched page falling below its normal range for several finalised days in a row, compared like for like with the same weekday over the previous eight weeks.
How CrawlWarden decides a Search Console number left its normal range 4 checks
The published method behind CrawlWarden's Google clicks and impressions alerts: same weekday medians over eight weeks, the thresholds, the floors below which nothing is judged, the confirmation counts, the estate wide date gate, and what the method cannot catch.
How often a healthy site answers Googlebot differently: our measurement 12 checks
Of 79 public domains healthy to a browser, 5.1 percent answered a Googlebot user agent differently from a German datacentre address. The full method, the raw counts and every limit on the number.
How to read these pages
Confirmation runs is how many consecutive observations of a new value are required before the change becomes an email. It is never less than two, and it is a property of the check rather than a setting: one observation is noise, and a deploy rolled back in three minutes should never reach anyone. Noisier checks sit at three or four.
Default is whether the check is on for a new site. 45 of the 68 are. The rest are available on every plan and off until you ask for them, because they are the ones that move on their own.
Scope is what a check reads. Page-scope checks run per watched URL. Site- and domain-scope checks run once per host however many URLs you watch there, cost almost nothing, and do not count against your URL limit.
No plan limits which checks you can run - the limits are on sites and URLs. See what each plan watches and what it costs.
Looking for the problem rather than the check?
These pages document the product. If you arrived looking for what a specific failure means and how to verify it yourself, the articles section covers each failure mode on its own page.