Soft 404 Detector

Google explicitly treats soft 404s as low-quality pages, not only does the page itself get deindexed, but Google extrapolates: "this site serves errors disguised as valid pages", which degrades the overall crawl budget and algorithmic trust. Search Console flags them in the "Soft 404 pages" report, one of the top KPIs to watch after a migration.

Most often: (1) removed product page that returns 200 with a "This product is no longer available" message, the CMS does not know how to serve a real 410. (2) Botched migration where old URLs return a SPA error page with 200 instead of a clean 404. (3) Automatic redirect to the home on any invalid path, the worst case for Google. (4) Under-construction page published too early, with lorem ipsum or placeholder content.

We combine 5 server-side signals: (1) HTTP status, it has to be 200 to qualify as soft 404. (2) Extremely short body, <50 words is highly suspect. (3) "Page not found" keywords in title/H1/start of body, FR + EN patterns. (4) URL contains 404/not-found. (5) Auto-redirect to home. Cumulated score ≥ 50 = near-certain soft 404. This heuristic catches ~90% of real cases.

Three patterns depending on the case. Definitively removed page: return HTTP 410 (Gone) with an explanatory body, Google deindexes quickly and stops re-crawling. Temporarily unavailable: HTTP 503 with Retry-After. Invalid URL typed by user: HTTP 404 with a real error page (similar page suggestions, search, return home). Never a 301 redirect to the home, that is disguised cloaking that pollutes signals.