Broken link checker

Crawl a site and find broken links, images and scripts, with the pages that link to each.

This tool needs a free account. Sign in or create one, then run it with your credits.

A page URL or a domain; a domain means its home page.

Pages to crawl on this site; costs 5 credits per 50.

Also check links to other sites.

Costs 5 credits per 50 pages (5 reserved for 50) with a free account.

What the broken link checker tells you

Links break all the time. Pages move during a redesign, files are deleted, and other sites close or reorganise. Each broken link is a dead end for a visitor and a wasted visit for a search engine, and a site with many of them looks neglected. They're hard to find by hand because nobody clicks every link on every page.

The broken link checker crawls your site from the page you give it, follows every internal link up to the page limit you choose, and checks every link, image, script and stylesheet it finds. You get a list of what's broken, grouped by kind, with the pages that link to each broken address and the link text, so you know exactly where to fix it.

How the crawl works

  • Start. We read robots.txt first. If it disallows the start page, we stop and you aren't charged.
  • Pages. We crawl breadth first, so the pages nearest your start page come first. A page counts as yours when it's on the same registrable domain, including subdomains such as shop.example.com. Each redirect is followed and recorded, so internal links that redirect are listed too.
  • Links. We collect links from anchors, images, responsive image sources, scripts and stylesheets. Email, phone and JavaScript links, links to other ports and links to a place on the same page are skipped and counted. Each address is checked once, however many pages link to it.
  • Other sites. Unless you turn it off, we check links to other sites too, obeying each site's robots.txt. We ask with a HEAD request first and fall back to a short GET when a server doesn't support HEAD.

What counts as broken

A link is broken when it answers 404, 410, another client error or a server error, or when its host name doesn't exist. Redirects that end on a working page are fine. A link that doesn't answer in time is listed separately as a timeout, because the site may only be slow. Links that need a login, that refuse automated checks, that robots.txt asks us not to check, or whose host was too busy to check are listed as unverified. Links to private or reserved addresses are never fetched and are reported as such.

Politeness and limits

The crawler fetches about one page a second from each host and at most two at a time, waits longer if robots.txt sets a crawl delay, and shares a limit of 120 requests a minute per host with every other DNS Super check. It crawls up to 500 pages and checks up to 5,000 links. A crawl that hits the time limit or that you cancel returns what it found so far, charged for the pages crawled, and isn't cached for anyone else.

Reading the result

The result opens with the number of pages crawled and links found broken, then lists broken links to your own pages, broken images, scripts and stylesheets, broken links to other sites, timeouts and unverified links. Each entry shows the status in words and opens to show the pages that link to it. Fix shared elements such as menus and footers first: one fix there often clears many entries.

Privacy

Crawls run under your account. Pages that aren't the site's home page are cached for your account only, so nobody else learns which addresses you checked, and the usage record keeps only the host name, never the full address you entered. Link text and referring pages appear in your result and nowhere else. We never send cookies or log in, so pages behind a login are listed as unverified rather than checked.

Questions

How is the price worked out?
5 credits per 50 pages, rounded up. The maximum for the page limit you choose is reserved when the crawl starts, and you pay only for the pages actually crawled. A crawl of a 20-page site with a 100-page limit reserves 10 credits and costs 5. Failed runs are free.
Do you obey robots.txt?
Yes. The crawler identifies itself as DNSSuperBot and follows robots.txt, including a crawl delay. If robots.txt disallows the start page, we don't crawl and you aren't charged. Links on other sites that robots.txt asks us not to check are listed as unverified.
Will the crawl slow my site down?
We fetch about one page a second from each host, at most two at a time, and at most 120 requests a minute to any one host across all our users. Link checks to other sites use a HEAD request first, which transfers no page content.
Why are some links listed as unverified?
Some pages need a login, some sites refuse automated checks with a 403 or a bot challenge, and some ask us to slow down. We can't tell whether those links work, so we list them separately instead of calling them broken.
How long does a crawl take?
About a second per page plus the link checks: a 50-page site usually finishes in two or three minutes. The crawl runs as a job, so you can leave the page and come back to the result.