Broken links

A broken link leads nowhere or produces an error. Yandex’s broken-link tooling separates them by whether anyone actually followed them, which is the split that turns a long list into a priority order.

The two categories

• Broken links without clicks — nobody tried to follow them. Caused by code errors, wrong links or deleted pages. Divided into external (from other sites to yours) and internal (within your own site).
• Clicks on broken links — real users followed them and reached an error page. This is the list with a measured audience.

Why some links never appear

Yandex explains the gaps, which prevents chasing phantom numbers:

• The page carrying the link is unknown to the Yandex bot.
• The link points to an alternate address of your site — the www or protocol variant.
• Where one resource links to the same page several times with identical link text, only one instance is shown.

So the report is a sample of what Yandex knows, not a complete inventory of every broken link pointing at the site.

How to work it

Clicked links first, because those errors have a measured audience and the frequency tells you which fix is worth doing. Then internal links, which are entirely within your control and cost nothing to fix. Then a judgement call on external ones: not every inbound broken link deserves a redirect, and the test is whether someone following it would find what they expected on the target page.

The maintenance argument

Broken internal links waste crawl attention, strand users, and — for external ones — leak the accumulated value of links other people chose to give you. A rising internal count is usually a signal that something in the publishing process broke, not that a few links need patching.