Broken Link Checker Free

The checker loads your starting page, collects every link on it, and tests each one. It then opens other pages on the same site that it found along the way and does the same, until it has read 10 pages or tested 80 different link targets, whichever comes first. Internal and external links share the 80.

Links are tested in the order they appear in the page source, which on most WordPress themes means the header menu first, then the content, then the sidebar and footer. On a homepage with a large mega menu, much of the 80-link budget can go on navigation before the checker reaches a single post.

So pick the start to match the question. To check a post’s outbound links, start on that post. To check the menus and footer, start on the homepage once; those links are the same on every page, so one pass covers them. To check a section, start on its category or hub page, since the pages it visits next are the ones linked earliest from there.

The crawl stays on the origin the starting URL lands on. If you type example.com and the site redirects to https://www.example.com, it crawls https://www.example.com. Links to any other host or protocol, such as an old http:// link on an https site, are tested but not crawled.

Pages crawled counts only pages that loaded. A link that appears in the menu, the content and the footer takes one place in the queue, and a page that fails to load or answers with an error is skipped, with a warning, so the next page in the queue takes its place. To keep the free check fast, it stops after 15 page requests, so a section full of dead pages can still end under 10. If the start page itself cannot be loaded, or is larger than about 900 KB, you get an error message instead of a report. If it loads but answers with an error status, its links are still tested and a warning gives the status.

Pages crawled, Links checked and Broken links are counts. Warnings tells you when a cap cut the run short: Link checks were capped at 80 links, or The crawl was capped at 10 pages. Either one means there are links this run did not test, so start again from a different page to cover more. A warning that some pages could not be loaded means the links on those pages were not tested; the pages themselves are in the table as broken targets.

The table lists every link tested, not only the broken ones, in the order they were checked. Broken targets are in red. Status is the HTTP code, or the error message when there was no code to show. Target is the link’s destination and Source is the page where it was found.

The report does not show the anchor text. To find a link on the source page, open that page, view its source and search for the target address. In the WordPress block editor, Options → Code editor shows the post’s markup, and your browser’s find will jump to the link.

Red does not always mean broken for your visitors. Some sites refuse automated requests while serving people normally. Open any doubtful link in a browser before you edit anything.

Illustrative values. A gardening blog checked a 2024 post that still gets search traffic. The run reported 10 pages crawled, 80 links checked and 5 broken links, plus the warning that link checks were capped at 80.

Three real problems and two false alarms. The 80-link cap also meant the footer of later pages was never reached, so a second run from the homepage covered the site-wide links.

Google has said for years that 404s on their own do not hurt how the rest of a site ranks. The cost of a broken link falls on readers who hit a dead end, and on links you control that no longer pass anyone anywhere. Work in this order:

A link that redirects is not reported, because it still works. If you want internal links to point straight at final URLs, check them separately in the Redirect Checker, or look at the internal link structure with the Internal Link Checker.

The Source column tells you the page, not the part of WordPress that printed the link. A link that appears on every crawled page is in a menu, a widget or a template; one that appears on a single page is in that page’s content.

When the same old address appears in dozens of posts, such as after a domain change, a search-and-replace beats editing by hand. The Better Search Replace plugin and WP-CLI’s wp search-replace command both offer a dry run that shows what would change; run it first, back up the database, and leave the guid column alone. Purge the page cache afterwards, or the checker will keep reading the old HTML.

This checker finds broken links on your pages. The opposite problem, other sites and old bookmarks linking to URLs you deleted, does not show up in any crawl of your own site.

Search Console lists those under Indexing → Pages → Not found (404). Inspecting one of those URLs shows, under Discovery, a referring page where Google found it, when one is known. On WordPress, a 404 log narrows it further: Tools → Redirection → 404s in the Redirection plugin, the 404 Monitor module in Rank Math, and 404 monitoring in SEOPress PRO.

Redirect the URLs that still get visits or have links from other sites to the closest live page with a 301. Leave the rest as 404s: a deleted page that nothing points to does not need a redirect, and redirecting everything to the homepage does more harm than good.

On a WordPress site connected to RankHive, the redirect side of this can come as drafts from its weekly Search Console review; each one waits for your approval, and its plugin serves the ones you accept.

Clear the page and CDN caches first, then repeat the run from the same starting URL so the same pages and links are tested. Rows you fixed should now be gone from the red list, and the Broken links count should only include the ones you decided to leave.

For an internal URL you redirected, run the old address through the Redirect Checker as well, so you know it lands on the new page in one hop. For URLs listed in Search Console’s Not found (404) report, use Validate fix only once every example URL either works or is meant to stay gone. Pages you deleted on purpose can stay listed there; Google rechecks them less and less often.

On a site that changes often, run the same three checks every few months: the homepage, the busiest post and the newest post.

Not every red row deserves an edit today. Bot blocks (403, 429 and 999) that open fine in a browser and a single 5xx that clears on a second run can be ignored. So can a 401 on a link meant for logged-in members.

Old comments with dead commenter websites, and posts that no one reads and that you plan to merge or retire, sit at the bottom of the list. If a post is about to be retired, redirect or delete the post rather than repairing its links.

The checker tests every link, including ones marked nofollow or sponsored, because a dead link frustrates a reader whatever its rel value. Those attributes change nothing about how urgent the fix is.