Case study: the SEO audit we ran on our own 393-page site

The site

393 static pages, English and Chinese twins under /en/ and /zh/, generated from a 68-film and 60-camera data set plus hand-written guides and tools.

What the first pass found

397 JSON-LD blocks, 43 invalid or missing across tools and dossiers.
One page with no meta description and several duplicate description strings.
15,152 internal links to check, plus 641 external links that could rot.
No lastmod and a broken sitemap pointer: robots.txt advertised /sitemap.xml, which did not exist.

What the crawler surfaced that a headless check would miss

Cloudflare Pages was answering HTTP 200 with the site root for every unknown path. /.env.production, /.git/config, /admin/login and any typo all returned the homepage. That is a soft-404: duplicate content for search engines, and wasted crawl budget on a new domain.

Once that was fixed with a real 404.html, the same fix exposed a second bug that had been hidden by the 200s: the sitemap pointer in robots.txt had been broken the whole time and nobody could see it.

Separately, www.openfilm.cc was serving the entire site as duplicate hostnames because a host-level redirect rule never fired.

What we changed

Generated JSON-LD for every tool, dossier and service page, then validated all 397 blocks.
Wrote a crawler that re-checks every internal and external link on each build.
Added a real 404 page and a robots prop so it can opt out of indexing.
Pointed robots.txt at sitemap-index.xml and added a 301 from /sitemap.xml.
Added a zone-level redirect for www to the apex domain.
Automated IndexNow so every new page is submitted to Bing and Yandex within 30 minutes of publication.

Verification

397/397 JSON-LD blocks valid.
15,152 internal links, 0 broken after the last build.
1 real 404 out of 641 external links, found and repaired.
392/392 sitemap URLs return HTTP 200 when fetched live.
Unknown paths now return 404 with a useful page instead of a 200 copy of the homepage.
Googlebot is crawling the content: 18 robots.txt fetches and dozens of article fetches in a 24-hour window, all 200.

What this means if you hire us

You get the same treatment: a crawl of your site, a prioritised list of what is actually wrong, the fixes written out in a form your developer can apply, and a second pass to prove each one landed. Scope is fixed at 150 USDC and delivery is five business days.