Methodology
How a sweep runs
We fetch what your server hands over, write down what we see, and score it against a published rubric. No agent to install, no code on your site.
- 1
You give us the domain.
That is all we need. No account, no crawl agent to install, no code on your site. If the report should be addressed to someone else, tell us and it will be.
- 2
We run the sweep.
We fetch every page the way an answer engine does, probe robots.txt, sitemap.xml, llms.txt and your structured data, check the apex over both http and https, and score six categories against a published rubric.
- 3
Senior staff reviews every finding.
Every run includes a Human in the Loop. A senior staff member checks the evidence, confirms the scores, and signs off before we put our name on the report.
- 4
You get two PDFs and a repair kit.
One report for the principal, one course corrections guide for whoever runs the site, and a zip of paste ready files. Every fix is ordered by impact against effort.
The method, stated plainly
Static fetch, no JavaScript
We request your pages the way GPTBot, ClaudeBot, PerplexityBot and CCBot do, and we read only what the server returns on the first pass.
Apex over both schemes
We check http and https at the apex and at www, and we record every redirect in the chain.
Evidence quoted verbatim
Every score carries the served bytes it came from. If you disagree with a judgment, the evidence is right beside it.
How repairs are ordered
Behavior happens when motivation, ability and a prompt arrive together. So the highest impact fix with the lowest effort goes first, and every item lands in one of three groups.
- Today, needs nobody's permission. Files you can publish this afternoon.
- This week, gated on a vendor, registrar, or the owner's own words. We write the ticket for you to send.
- Not yours to write. The owner's bio, claims and credentials. Nobody else can supply those honestly.