What it is.
A crawler reports; the analysis is the work. This walks the pipeline from crawl export into Claude Code, with the instruction file we include on every project.
Pattern-level issues account for roughly 65% of the ranking-impacting findings in the B2B SaaS audits we run. The remaining 35% are individual pages, and most resolve once the template pattern is corrected.
About 70 to 80% of what Claude surfaces passes human verification unchanged, and the rest gets reshaped by the lead, usually on prioritization.
What you get.
- The pipeline from crawl export through to Claude Code
- The audit instruction file we include with every project
- Five prompts covering orphan pages, redirect chains, duplicate patterns, crawl depth and schema gaps
- The pattern-versus-page split, at roughly 65% and 35% of ranking-impacting findings
- A verification pass rate of 70 to 80% on surfaced findings
- One audit walked end to end on a dated timeline, plus the three checks that stay manual
Read it at the source.
Read the guide
How much of the output needs checking?
Do I need Screaming Frog?
Can AI run a technical SEO audit?
Why cross-reference Search Console in an audit?
What share of audit findings are template-level?
How much AI audit output can you trust?