For years, broken links were a well-understood problem with a straightforward consequence: a bad user experience and a small dent in your search rankings. Fix them periodically, move on. Most site owners treated link audits like changing a car's air filter — important, but easy to put off.
That calculus changed in 2024, and by 2026 the stakes are significantly higher. The search landscape has shifted dramatically. Google now surfaces AI-generated answer boxes — called AI Overviews — at the top of results for a huge portion of queries. ChatGPT has its own web search. Perplexity is a go-to research tool for millions of users. All of these systems are crawling the web, pulling from source pages, and deciding which sites are credible enough to cite.
Broken links are now a trust signal. And a site with a broken technical foundation is increasingly one that AI search systems quietly pass over.
How AI search actually crawls and evaluates your site
To understand why broken links matter more now, you need to understand what AI-powered search engines are actually doing when they crawl your content — because it's meaningfully different from how traditional crawlers worked.
Traditional Google search used to be primarily about keywords and links: does this page contain the right words, and do enough authoritative pages link to it? That still matters. But AI-generated answers require a much higher level of confidence in a source. These systems don't just check that a page exists — they evaluate whether the page is healthy, complete, and trustworthy enough to be cited in an answer that will appear in front of millions of users with no further disclaimer.
Think of it this way: a traditional search result links to your page and lets the user judge for themselves. An AI Overview, a ChatGPT answer, or a Perplexity citation presents your information as fact. The system has to be confident you're a reliable source before it stakes its credibility on you.
Traditional search: "does this page rank well?" — AI search: "is this site trustworthy enough to quote without embarrassing us?" These are very different standards, and technical site health is a much bigger factor in the second question.
Three AI crawlers, one broken site
It's not just Google anymore. Your site is being evaluated by multiple AI systems simultaneously, each with its own crawler and its own quality signals.
What all three have in common: they want to cite pages that are stable, complete, and well-maintained. A page that returns working content with no broken links or missing assets signals a reliable, well-managed source. A page riddled with 404 errors, broken images, and missing CSS sends the opposite message.
What signals do AI crawlers actually use?
AI search systems don't publish their exact ranking criteria, but based on how Google's AI Overviews have evolved, published patents, and the stated principles behind systems like Perplexity, we can identify the signals that consistently surface authoritative, citable sources:
| Signal | What it means for AI search | Broken links impact |
|---|---|---|
| Page completeness | Does the page load fully, with all assets present? | Direct — broken images & CSS |
| Internal link integrity | Are internal links pointing to real, live pages? | Direct — 404 internal links |
| E-E-A-T signals | Experience, Expertise, Authoritativeness, Trust | Indirect — affects trust score |
| Page load speed | Can the crawler access content efficiently? | Indirect — redirect chains add latency |
| Content freshness | Is the site actively maintained? | Indirect — broken links signal neglect |
| External link health | Do outbound citations point to live sources? | Direct — broken external links |
| Crawl accessibility | Can the crawler index key pages efficiently? | Improved by fixing all above |
Notice that broken links appear as a direct negative factor in three separate signal categories. Fix broken links and you're not just cleaning up one metric — you're improving your standing across multiple dimensions that AI search systems use to assess your credibility.
The trust problem: broken links as a quality signal
Here's the thing about broken links that becomes especially relevant in an AI-first world: they're not just technically problematic. They communicate something about how you manage your site.
A site with multiple broken internal links is telling crawlers — and the AI systems that depend on them — that nobody is actively maintaining the content. Pages were moved or deleted without updating the links that pointed to them. Outbound citations were never checked after being published. Images were removed without replacing them. These are the hallmarks of a site where the content was written and then abandoned.
AI systems are specifically trying to avoid citing sources that might be outdated, unreliable, or poorly maintained. From their perspective, an actively maintained site with clean internal links and working external citations is far safer to quote than one that shows signs of neglect.
Every broken link on your site makes you slightly less likely to be cited in an AI-generated answer. On a site with dozens or hundreds of broken links, this compounds into a meaningful disadvantage against competitors who have cleaner technical foundations — even if their content isn't as good as yours.
What's changed since 2024 specifically
It's worth being precise about what's new here, because the basic advice ("fix your broken links") isn't new. What's changed is the scale and mechanism of the consequences.
AI Overviews now appear in roughly 15–20% of searches
That number has grown significantly since Google rolled out AI Overviews globally. For informational queries — exactly the kind of content most blogs and informational sites publish — the AI Overview often takes the entire above-the-fold space. If you're not in it, you may not get the click at all, even if you rank #1 in the traditional results below it.
ChatGPT Search is sending real traffic
OpenAI's web search has matured from a novelty into a tool that millions of people use daily for research. ChatGPT now cites sources directly, and those citations drive referral traffic. Sites that are indexed and trusted by OAI-SearchBot are getting a new traffic stream that didn't exist two years ago. Sites with poor technical health are invisible to it.
Perplexity is used by a highly valuable audience
Perplexity's user base skews heavily toward professionals, researchers, and decision-makers — exactly the kind of high-value audience most sites want to reach. Getting cited by Perplexity isn't just about raw traffic volume; it's about the quality of the audience that follows the citation.
What to actually fix — prioritised by AI search impact
Not all broken links are equally damaging to your AI search visibility. Here's how to triage your fixes based on impact:
Priority 1 — Broken links on your most-cited pages
Which pages on your site get linked to most from external sources? Which ones rank best in traditional search? Start there. These are the pages AI systems are most likely to evaluate as potential citation sources. A broken image or a dead internal link on your best-performing page is disproportionately damaging.
Priority 2 — Broken external citations
If your articles link out to external sources as references or citations, and those sources have gone offline, you're now pointing your readers (and crawlers) to dead ends. AI systems that are specifically evaluating whether your content is well-sourced will flag this. Go through your reference links and either update them to live URLs or remove them.
Priority 3 — Missing images on key content pages
A page with missing images isn't just visually broken — it signals to a crawler that the page's content is incomplete. For AI systems trying to extract information from your page, a missing image placeholder where an infographic, chart, or diagram used to be suggests the page may have been partially stripped of its content.
Priority 4 — Broken internal links sitewide
After high-priority pages, audit your full site for internal links pointing to 404 pages. These waste crawl budget, dilute PageRank, and contribute to the overall signal that your site is poorly maintained. A full-site scan with a tool like Broken Link Checker will surface all of them in one pass.
Run a scan filtered to your top 20 most important pages first. Fix any broken links on those pages this week, before touching anything else. You'll get the highest AI search impact for the smallest amount of work.
The E-E-A-T connection
Google's framework for evaluating content quality — E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) — has been explicitly extended to how it evaluates sources for AI Overviews. The "T" in E-E-A-T, Trustworthiness, is particularly relevant here.
A site that Google (and by extension its AI systems) considers trustworthy is one where the content is accurate, the links work, the sources are cited properly, and the site is clearly being actively maintained. Broken links are a direct signal against that last criterion. They're the digital equivalent of a business with a broken sign, a dead phone number on the website, and an outdated address on Google Maps — technically still present, but clearly not well looked after.
Improving your E-E-A-T signals is a longer project that involves content quality, authorship, and backlink building. But fixing your broken links is one of the fastest, most concrete technical steps you can take right now — and it addresses the Trustworthiness component directly.
What a well-maintained site looks like vs a neglected one
- Multiple internal links returning 404
- Missing product or article images
- Outbound citations pointing to dead pages
- Redirect chains on important URLs
- Pages not updated in 2+ years
- CSS or JS files failing to load
- All internal links resolve to live pages
- Images load correctly across all pages
- External citations checked and updated regularly
- Single-hop redirects only, no chains
- Content actively reviewed and refreshed
- All assets (CSS, JS, fonts) load cleanly
Your action checklist for AI search readiness
The bottom line
Broken links have always been a bad thing. In 2026, they're a worse thing — because the consequences now extend beyond traditional search rankings into whether AI systems consider your site reliable enough to cite.
The good news is that this is one of the most actionable technical SEO improvements you can make. You don't need to wait for content to rank or for backlinks to accumulate. A scan, a few hours of fixes, and a regular audit schedule puts you in a fundamentally stronger position than the majority of sites that are still treating broken links as a low-priority housekeeping task.
AI search is not a future trend. It's the current reality that's shaping which sites get visibility and which ones get passed over. The sites that maintain clean, well-structured, fully working pages are the ones AI systems trust — and trust, in 2026, is the currency that determines whether your content reaches readers at all.