Docs › Keeping a site healthy
Link Doctor scans every published post and page for broken hyperlinks — internal and external, but never Amazon product links, which the Product Health Monitor owns. Version 1 of this tool flagged live websites as dead; version 2 exists because of that mistake.
In one lineA link is reported dead only on unambiguous proof — a 404/410, or a domain that fails to respond twice in a row. Anything softer counts as alive.
Why “dead” got stricter
The first version was too eager. It flagged live external sites as dead because it treated any non-200 response as proof of death. Bot-walls (403/405), rate limits (429), momentary server hiccups (5xx) and Cloudflare challenge pages all produce a non-200 response from a site a human visitor would still reach perfectly well. Reporting those as dead sent owners removing links that worked.
The current rule reports a link dead only when the evidence is unambiguous:
- A final HTTP status of 404 or 410, after following redirects — and a 404 is confirmed with a second, independent request before it is reported.
- Or the domain fails to respond at all, twice in a row.
Everything else — a bot-wall, a rate limit, a temporary 5xx, a security challenge page — is treated as alive. Checks use full browser-style headers and a real GET request rather than a HEAD, because many servers give HEAD requests a different (and less reliable) answer than a real visitor would get.
What it never touches
Amazon product links, Amazon’s short-link domains (amzn., a.co) and any non-http link (mailto:, anchors, tel:) are excluded automatically. Amazon links are the Product Health Monitor’s job specifically, because verifying an Amazon product needs the Amazon API, not an HTTP status code — a discontinued product can still return a working page.
Removing a dead link only unwraps it
The fix is deliberately narrow: removing a dead link strips the <a> tag and keeps the anchor text in place. Nothing else about the sentence changes, and no surrounding content is touched. Up to 200 links are removed per click, which is a server-safety limit rather than a real ceiling — select more than that and clicking Remove again clears the next batch.
Results persist between visits
A scan’s results are saved server-side, so leaving the tab and coming back later still shows the same report rather than forcing a re-scan. Results stay in sync as you remove links, so the count you see always matches what is actually still on the site.
Common questions
Why is a link I know is broken not showing up?
It may be responding with something other than a clean 404/410 — a redirect to a working page, or a soft-404 the server answers with a 200. Both count as alive under the strict rule, deliberately, to avoid the false positives version 1 had.
Why did a link that returns 403 not get flagged?
A 403 usually means the site is blocking automated requests, not that the page is gone. A human clicking the same link would typically still reach it, so it is treated as alive.
Does this check Amazon product links too?
No, they are excluded on purpose. Product Health Monitor checks those against the Amazon API, which is the correct way to verify a product rather than just its URL.
Will removing a dead link change anything else in the post?
No. Only the anchor tag is unwrapped; the visible text stays exactly where it was.
Next
- Product Health Monitor — for Amazon links specifically.
- The 18-point SEO audit — dead links feed directly into it.
- Finding dead Amazon links your link checker cannot see.