Broken Link Monitor

Find broken links on your own website. Each broken link is one row: the link, the page it was found on, the link text, the status, and whether it stays on your site.

General information only.

Links that a browser adds with JavaScript are not seen. A link that works in your browser can still fail from a data center.

What you get

For each public site:

• Broken links. A link is broken when the target returns an error status, the name does not resolve, the connection fails, or the request times out.
• The page the link was found on, and the link text (shortened).
• Whether the link stays on your site or points somewhere else.
• Redirect chains of more than two hops.
• Plain http links found on an https page.
• In monitor mode, which broken links are new since the last run and which are fixed.
• One summary row per site, even when no link is broken. That row has the site address and the counts only.

Pages that do not answer are not charged. Private addresses, internal names, and URLs on a port other than 80 or 443 are skipped and are not charged. The summary counts every URL you supplied, how many sites were accepted, and how many were skipped. A run whose addresses are all private still finishes, with nothing crawled.

If robots.txt answers with a server error, that site is not crawled. If the site does not answer at all, the start page is reported as broken and is not charged.

Tutorial

1. Paste your own site URLs under Website URLs. Use a full URL such as https://example.com/. Up to 20 sites. Extra sites beyond that cap are counted and not checked.
2. Set Pages per site. The default is 200 and the maximum is 2000. The crawl starts at the URLs you gave and also reads sitemap.xml when the site publishes one.
3. Leave Check links to other sites on to check each external link once. Turn it off to stay inside your site. Other sites are never crawled.
4. Start the run. Open the dataset. Each site has one summary row, plus one row for each problem link. The OUTPUT record has the run counts.
5. To watch the site, turn on Monitor mode, keep the store name, and save the input. Schedule the Actor daily or weekly. The next run marks broken links as new or still, and adds a row with fixed when a link from the last run works again. The first monitored run marks every broken link as new.

mailto and telephone links are skipped. Query strings are not checked or saved.

Pricing

You pay a small fee when the run starts, plus a fee for each page that answered. A page that does not answer, or a private address, is free. Checking a link is included in the page fee. Rows in the dataset are free.

Event	Price
Run start	$0.00005
Page crawled	$0.0005

What is not collected

Page text is not saved. Image alt text is not stored. Email addresses and phone numbers are not saved. Link text that is a contact request is not saved. Query strings and fragments are not saved. Logs name the domain, how many links were broken, and how many pages answered.

Limits

Up to 20 sites in one run, and up to 2000 pages on each site. There is a limit on how many links one run will check. Links past that limit are counted and not checked. Each request times out, redirects are capped, and response bodies are capped. Only http and https on ports 80 and 443 are accepted.