Canonical Signal Audit: When Six Pages Pointed to the Homepage
A measured baseline showing how a site-wide canonical default obscured the identity of otherwise distinct pages.
Inspect the evidenceFIELD LAB
Measured findings, including the ones that exposed our own mistakes.
These reports start with a reproducible observation, publish the raw snapshot, state what the evidence cannot prove, and record the resulting change.
10 REPRODUCIBLE CHECKS
Evidence before recommendations.
Seven-Page Server Response Baseline From the Hermes VPS
A reproducible one-request baseline for response time and HTML transfer size across the site's main templates.
Inspect the evidenceCache Header Field Check Across HTML and Discovery Files
An observed comparison of cache signals on HTML, robots.txt, sitemap.xml, and ads.txt before remediation.
Inspect the evidenceSecurity Header Baseline for Seven Public Page Templates
A transparent header inventory that separates observed omissions from claims about exploitability.
Inspect the evidenceContent Depth Comparison: Indexes, Tools, Guides, and Articles
Observed word and heading counts used to decide which templates should remain indexable during editorial recovery.
Inspect the evidenceInternal Link Discovery Map for the Main Site Templates
A server-rendered link inventory used to rebuild navigation around evidence rather than content volume.
Inspect the evidenceRobots, Sitemap, and Ads.txt: Three Files With Different Jobs
A direct response check of the site's machine-readable discovery and advertising declarations.
Inspect the evidenceRedirect and Hostname Baseline: HTTP Worked, WWW Did Not Resolve
A two-entry host normalization check that records redirects and DNS failure without guessing at crawler behavior.
Inspect the evidenceThird-Party Script and Form Inventory Before Advertising
A bounded privacy baseline of script origins and server-rendered forms before any advertising code is enabled.
Inspect the evidenceThe Missing Visual Evidence Problem: An Accessibility-Aware Baseline
Why zero missing alt attributes was not a success metric when the sampled publication also contained zero images.
Inspect the evidence