You've probably seen the line before: Google officially treats 404 errors as a normal part of the web, not a penalty. But there's a big gap between "it doesn't directly hurt rankings" and "it has no impact at all", especially now that AI answer engines (ChatGPT, Perplexity, Google AI Overviews) run their own validity checks on the links they're about to cite. A dead link today isn't just a bad user experience anymore. It's a broken freshness signal, read differently by a classic crawler than by a generative engine.
This guide covers both angles: what Google actually says about the SEO impact of 404s, why the topic gets more sensitive on the GEO side, and the concrete method to audit, fix, and prevent broken links on a SaaS or e-commerce site.
404, 410, soft 404: the distinction most people get wrong
Three status codes get constantly mixed up, even though they send very different signals to Google:
- 404 (Not Found): the page doesn't exist at this URL, but it might be temporary. Google keeps checking back on it periodically.
- 410 (Gone): the page has been permanently removed and won't come back. Google typically deindexes faster than with a 404 and stops repeatedly crawling the URL.
- Soft 404: the worst of the three. The page returns an HTTP 200 (OK) status but displays "not found" content, an empty cart, a no-results internal search, a delisted product page. Google detects the pattern and flags it as a soft 404 in Search Console, but since the server never explicitly announces it, Googlebot keeps crawling it like a normal page, wasting crawl budget for nothing.
On e-commerce sites, soft 404s are by far the most common of the three and the least monitored: a permanently out-of-stock product page that stays live with a simple "unavailable" banner instead of a real 410 happens on nearly every catalog with more than a few hundred SKUs.
What Google actually says about the SEO impact
John Mueller (Google) has repeated this multiple times: having 404 or 410 pages is normal, even at scale, and it doesn't lower the ranking of the rest of the site. No direct penalty, then, but three real indirect effects:
- Loss of accumulated authority. If an old page had quality backlinks and you remove it without a redirect, all the accumulated SEO equity disappears with it. No one "punishes" your site, but you're throwing away an asset that took months to build.
- Wasted crawl budget. Every crawled 404 or soft 404 consumes a Googlebot pass for zero benefit. On a catalog with thousands of URLs, that can eat a meaningful share of the total budget, we break down the full mechanics in our crawl budget guide.
- Degraded user experience. A visitor who clicks a search result and lands on a dead page usually bounces to a competitor, and that behavior (a quick return to the results page) is a signal Google observes indirectly through aggregated click data.
The GEO angle: why a dead link can cost you an AI citation
This is the part most general SEO guides still don't cover, and it's where things get interesting for an audit focused on AI visibility. Answer engines like ChatGPT, Perplexity, or Google AI Overviews don't just index a page once and forget it, several of them run a freshness/validity pass on the sources they're about to cite in an answer. Concretely: if your page links out to a study, source, or resource that now returns a 404, the engine has a measurable reason to demote your page in the citation ranking for that specific query.
Two direct consequences for a site that wants to get cited:
- Dead outbound links hurt the citing page's credibility. A page linking to 5 sources where 2 are dead sends a signal of neglected maintenance, the exact opposite of what an engine is looking for when choosing a reliable source to cite.
- Internal 404s create holes in the site's knowledge graph. An AI crawler following a broken internal link loses track of the content hierarchy, a problem directly tied to what we cover in our internal linking guide.
In other words: a broken-link audit is no longer just routine SEO maintenance. It's a prerequisite for staying citable.
How to audit your broken links (4-step method)
- Full site crawl. Use Google Search Console (the "Pages" report, filtered on 404 errors) for a free first pass, then run a third-party crawler to also catch internal broken links Google hasn't encountered yet.
- Check your outbound links. Internal links show up in a standard crawl, but dead outbound links (to sources, studies, third-party tools) need a dedicated check, often the blind spot of an SEO audit that only looks at the site itself.
- Cross-reference with inbound backlinks. Before deleting or letting a page die, check whether it still receives active backlinks. A page with low direct traffic but 10 quality backlinks deserves a redirect, not abandonment.
- Prioritize by impact. Not all 404s are equal. Rank them by historical traffic lost, number of inbound backlinks, and depth in the internal link graph, fix what costs the most first.
On a typical audit run through SeAudit, a mid-sized e-commerce site (600-800 pages) usually turns up around thirty internal broken links and a dozen dead outbound links to third-party resources, most of them invisible to the naked eye as long as the site keeps looking like it's working fine on the surface.
The decision tree: 410, 301, or leave it alone
| Situation | Recommended action |
|---|---|
| Page permanently deleted, no traffic or backlinks | Leave as 404 (or switch to 410 to speed up deindexing) |
| Page deleted but still has traffic or active backlinks | 301 redirect to the closest relevant page |
| Page moved or renamed | Direct 301 redirect to the new URL (never chain A → B → C) |
| Product temporarily out of stock | Keep the page live (status 200), clearly show the stock status, suggest alternatives |
| Product permanently discontinued | 301 redirect to the parent category, or switch to 410 if no equivalent page exists |
One simple rule to remember: never default to redirecting everything to the homepage. An irrelevant redirect (product page → homepage) is read by Google as a low-quality signal almost as negative as an unhandled 404, and it's frustrating for a user who was looking for something specific.
Preventing new broken links from appearing
A one-off audit fixes what's already broken, but without a process, the problem comes back every time content or the catalog gets updated:
- Stable URLs by design. Avoid tying a URL to an attribute that will change (price, stock status, date). A URL that never moves never needs a redirect.
- Systematic redirects on every page deletion or rename, built into the editorial or catalog workflow as a mandatory step, not an afterthought.
- Quarterly audits at minimum, monthly if the catalog changes often (new products, blog posts, removed seasonal promotions).
- Check outbound links before publishing any article citing external sources, a link that works today can be dead in six months, and the fact it worked at writing time doesn't mean it still will.
FAQ
Do 404 errors really lower my Google ranking?
No, not directly, Google has confirmed this repeatedly. The impact is indirect: lost authority if a page with backlinks disappears without a redirect, wasted crawl budget, and a degraded user experience.
Should I redirect all my 404s to the homepage?
No. An irrelevant redirect is nearly as bad as an unhandled 404 in Google's and users' eyes. Redirect to the closest relevant page, or leave it as a 404/410 if no alternative exists.
What's the practical difference between 404 and 410?
A 404 signals a potentially temporary absence, a 410 a deliberate, permanent removal. Google typically deindexes a 410 faster, which limits crawl budget waste on pages that will never come back.
Can a dead link in my content really keep me from being cited by ChatGPT or Perplexity?
It won't automatically exclude you, but several AI answer engines run a freshness check on sources before citing them. A dead outbound link is a signal of neglected maintenance that puts you at a disadvantage against a better-maintained competing source.
Key takeaways
- A single 404 isn't an SEO disaster, it's the unmanaged accumulation that becomes a problem (crawl budget, lost authority, user experience).
- Always distinguish between 404, 410, and soft 404: they're three different signals for Googlebot.
- The real 2026 angle is GEO: AI engines check link validity before citing a source, internal or external.
- Prioritize fixes by actual impact (traffic, backlinks, link depth), not alphabetical URL order.
Want to know how many broken links are dragging your own site down, and which ones to fix first? Get your free /100 score in a few minutes, or grab the full PDF report for a detailed audit with a prioritized action plan.