A dead link tells a crawler something before a person even notices it.
I find them on almost every audit. An old blog post points to a service page that got deleted. A PDF got pulled when a host changed. A page moved and nobody built the redirect. The link still sits there, pointing at nothing.
People skim past a 404 and move on.
What reads the page next does not.
It logs the dead end and quietly downgrades what it expects to find on the rest of the site.
What a 404 Actually Signals
A single broken link is not a crisis. A pattern of them is a signal. When search bots and AI tools hit a dead end on page after page, they start treating the whole site as less reliable, not just the one link.
That matters more for AI search than it ever did for a ranking list. An AI tool deciding whether to cite your site as the source for a fact is weighing how current and well kept it looks. A site full of dead ends reads as unmaintained, and unmaintained sites do not get picked as the source behind an answer in ChatGPT or Perplexity.
I see this most on sites that have been live for years without anyone checking the plumbing. The homepage looks fine. Three folders deep, it is a different story.
Where They Pile Up
Dead links collect quietly, usually in the same few places. Old blog posts that link forward to a service you renamed. A navigation menu that still points to a page you killed last year. External links to a resource another company took down.
Service businesses around Phoenix are some of the worst offenders I check, and not because anyone was careless. A site redesign swaps page paths without anyone checking what pointed to the old ones. Our post on what a site migration can break covers the bigger version of this. The slow version, the one nobody schedules a check for, does the same damage over months instead of overnight.
Redirect Chains Make It Worse
A single redirect from an old URL to its replacement is fine. A redirect that points to another redirect that points to a third one is not.
Crawlers follow a limited number of hops before giving up, and every extra hop in the chain is a chance it stops before it reaches the live page.
I see three-hop chains on sites that have moved their content management system twice. Each move added a layer instead of replacing the one underneath it. The fix is not clever. Point every old URL straight at wherever the content actually lives now, in one hop.
What This Costs in Citations
Schema often references the same URLs the broken links point to. A review marked up with it that links to a since-deleted product page is now vouching for nothing. The record coming apart in small pieces like this is part of why our post on entity consistency matters here too.
An AI tool weighing whether to recommend a business pulls from whichever source looks complete and current. A site with working links and a thin one with broken links can carry the same information.
The thin, working one gets cited more.
That is not a close call. It is the plainest version of the choice AI search makes every time it picks a source.
The Fix Is Maintenance, Not a Rebuild
This is not a reason to rebuild a site. It is a reason to check it on a schedule instead of only after something breaks. A quarterly crawl that lists every 404 and every redirect chain catches most of the damage before it compounds.
Fix the pages with the most incoming links first. A dead link buried three folders deep with nothing pointing to it matters less than one sitting in your main navigation or inside last month's highest-traffic post.
I run this kind of check as part of every SEO / AEO / GEO Foundation engagement, not as a one-time favor before launch.
A site that is technically fine but quietly full of dead ends is still losing ground with every crawl. Checking for it costs almost nothing. Not checking costs more than it looks like from the outside.
If you want that audit run properly, the SEO / AEO / GEO Foundation is where that work lives.