Broken links are hyperlinks whose destination no longer returns a useful, reachable resource, creating a failed path for people and crawlers.
A broken link can be an internal URL to deleted content, an external reference that moved, a malformed relative path, or a target that now returns a server error. It is a navigation defect, not simply a URL that looks old.
Practical implementation
Crawl a defined page set and group failures by final response and referring URL. Repair a link directly when the resource still exists, use a relevant permanent redirect when a replacement exists, restore a needed page, or retain a true 404 when no equivalent is available. A home-page redirect is not a universal repair.
What to verify
Follow the rendered link from its source, record the final status and target, then re-crawl after the change. Check that the source link, sitemap and internal navigation agree. Monitor external references because their targets can change again.
Common mistakes
Do not convert a crawl count into a ranking claim. Prioritise paths by user importance and actual availability of a relevant destination.