Broken links (404) in a scan report

In your report

In a finished scan report, Broken links (404) is a card in the overview and a filter in the findings list. It collects the outbound links on your pages whose destination answered with 404 (Not Found) or 410 (Gone). In other words, the server was reached, and it said the page is not there.

The card shows four numbers: Broken or gone, Server errors, Unreachable and Couldn’t check. Only the first is the list this page is about. The other three are shown beside it so that nothing is hidden, but they are different situations, covered below.

In the findings list, each broken link carries a 4 Broken (404) priority chip and a 404 or 410 chip showing the status code. The card itself is labelled Priority 4 in the standard order. The order can be changed in your settings, so the number on your own card may differ.

What counts as broken, and what does not

Before a link is called broken, Link Signals asks again. Many sites answer an unknown crawler with an error that a normal browser would not get, so a failed request is retried once with a browser-style user agent, and the better answer is kept. Only a 404 or 410 that survives that second try is reported as broken.

A Couldn’t check link is not a broken link. The site answered, it simply declined to show the crawler a page, so open it yourself if you want to be sure. The same applies to unreachable links: a site that was down for an hour during the scan is not the same as a page that has been removed.

Why it matters

• Visitors who click a broken link hit a dead end, which reflects on the page they were reading.
• A page full of dead links looks neglected, to people and to anyone judging the quality of the page.
• Links to pages that have gone are often links to businesses or sites that have gone too. When the whole domain lapses, anyone can register it and answer your visitors with whatever they like.

That last point is why broken links sit below some other findings. A link to a domain that is parked or not registered is a risk right now, so those rank first and second. A 404 on a domain that still exists is an inconvenience. It ranks fourth, after links that are concealed or unrelated to the page, and ahead of the general Needs review list. See finding priorities for the full order.

If a broken link’s domain has also expired or is not registered, the finding moves up to that higher priority. It is counted once, in the highest one it matches.

What to do

1. Open the link and the page it sits onUse the Found on column to see the page that carries the link, then click the link as a visitor would. Some destinations return a 404 to automated requests and a normal page to people, so confirm it fails in a browser.
2. Check whether the page has movedGo to the destination’s home page or use its search. Sites often restructure, and the same article lives on at a new address. If you find it, replace the link with the new address.
3. Look for a good replacementIf the page is truly gone, search for another source that serves the same purpose, or an archived copy you are happy to cite.
4. Or remove the linkIf nothing replaces it, delete the link and keep the sentence. Rewrite the sentence if it only made sense with the link.
5. Redirect your own old pagesIf the broken link is to one of your own pages, a link pointing at a page you deleted, restore the page or add a redirect from the old address to the closest current one.
6. Check how the link got thereA single stale link is normal ageing. A page with many broken links to the same unfamiliar domain is worth a closer look at who added them.

Common mistakes

• Treating Couldn’t check as broken. A 403 or a bot check is not a missing page. Removing a good link because a site refuses crawlers loses a useful reference.
• Pointing everything at the home page. A broken link to a specific article is better replaced by an equivalent article than by a generic home page.
• Ignoring unreachable links indefinitely. One failed request may be a bad moment, but a link that stays unreachable across scans is probably gone for good.
• Fixing only the first page. The same link is often repeated in a footer, a template or an old post. The finding lists the pages it was found on.

How to confirm the fix

Once a link has been changed or removed, use Re-check on the finding to test it again without running a whole scan. A link that now returns a normal page, or is no longer on the page, stops being reported. If you have decided to leave a link as it is, for example one that works only for logged-in users, use Mark reviewed so the decision is recorded and the finding is checked again on the next scan.

How this counts in your LinkSignal

Links that return 404 or 410 are counted in the broken links row of your site’s external LinkSignal. Links that could not be reached, were blocked or returned a server error are not in this list and are not scored. A broken link that is also parked or unregistered is counted once, in the row that comes first in the default order. The LinkSignal overview lists every row, and how the score is worked out gives the points. LinkSignal is Link Signals’ own metric, worked out only from what this tool finds, and it is not connected to any search engine or third-party score.