📄 Trang

404 Errors: Which Ones Actually Matter

A 404 error is not automatically an SEO emergency. Most are routine requests for URLs that never existed, expired campaign links, mistyped addresses, or bots probing for files; fixing every...

📅 Cập nhật 18/09/2026 10 phút đọc

A 404 error is not automatically an SEO emergency. Most are routine requests for URLs that never existed, expired campaign links, mistyped addresses, or bots probing for files; fixing every one can waste days and create worse indexing problems.

The 404 errors that matter are the ones attached to value: pages with backlinks, meaningful historic traffic, current internal links, conversions, or a legitimate replacement page. The practical job is not to eliminate every 404 error. It is to separate lost assets from harmless noise.

What a 404 error actually tells search engines

A 404 error is an HTTP response meaning the server cannot find the requested resource. It is usually the correct response when a page has been permanently removed and there is no relevant alternative.

Search engines expect 404s. Sites change, products are discontinued, URLs are mistyped, and external sites retain old links. A normal 404 response helps crawlers understand that a URL should not be indexed. If a page remains unavailable, search engines will generally recrawl it over time and remove it from search results.

The important distinction is between a real missing page and a page that only looks missing. A real 404 returns the HTTP status code 404 or, in some cases, 410 Gone. A soft 404 returns 200 OK but gives users an empty, generic, thin, or “not found” page. That mismatch can make crawling and indexing less efficient because the server says the page is valid while the content says otherwise.

Do not judge status codes by what appears in the browser alone. A branded error page can look polished and still return 200, which is a technical mistake. Check the actual response in server logs, a crawl, browser developer tools, or an HTTP header checker.

Which 404 errors matter: a practical triage table

Use this table before creating redirects. Start with the URL’s value, then decide whether there is a genuinely equivalent destination.

404 type Does it matter? Typical signal Recommended action
Deleted page with relevant inbound links Yes, usually high priority Referring domains, editorial links, or links from partner sites Restore the page if appropriate, or use a 301 redirect to the closest equivalent page
Former high-traffic landing page Yes Organic sessions, paid campaign visits, conversions, or assisted revenue in the previous 12 months Rebuild, redirect to a close substitute, and update internal links and campaigns
Broken internal link to a removed URL Yes Site crawl finds links from navigation, articles, category pages, or XML sitemaps Update or remove the internal link; redirect only where a relevant replacement exists
Typo in an external link with no traffic or links of value Usually no One-off URL, zero visits, no meaningful backlinks Leave as a true 404, or contact the linking site if the relationship warrants it
Bot request for /.env, /wp-login.php, or random scripts No SEO priority Unrecognizable paths, repeated automated requests, security scanning patterns Keep returning 404 or 403 as appropriate; handle through security monitoring, not SEO redirects
Expired product with no successor Sometimes Product URL has links, traffic, or appears in internal navigation Use a relevant category page only if it helps the shopper; otherwise return 404 or 410 and remove links
Soft 404 returning 200 Yes “Page not found” copy, sparse template, or empty search results with a 200 status Return a real 404 or 410; improve the error-page experience separately

Prioritize by inbound links, traffic history, and business value

The fastest way to turn a long 404 report into a useful worklist is to score URLs by evidence of lost value. Search Console, analytics, crawl data, server logs, and backlink tools each reveal a different part of the picture.

1. Check inbound links first

A missing URL with legitimate inbound links deserves attention, especially if the links come from relevant publications, suppliers, universities, associations, customers, or well-maintained industry resources. One highly relevant editorial link can be more valuable than dozens of low-quality scraper links.

Create a high-priority bucket for URLs with at least one credible referring domain and a clear replacement target. Review links manually before redirecting. If the old page was a guide to payroll software and the closest current page is a payroll software guide, a 301 redirect makes sense. If the only available destination is a generic homepage, it usually does not.

Be cautious with legacy links to obsolete content. If a discontinued report, event, or product has no comparable successor, returning a proper 404 can be more honest than forcing visitors onto an unrelated page.

2. Look at the last 12 months of traffic

Traffic history helps distinguish a theoretically valuable URL from one that still affects users. Review at least 12 months where possible, so seasonal peaks do not disappear from the analysis. For businesses with long buying cycles, review 24 months as well.

Prioritize a 404 page when it had any of the following before it disappeared:

  • More than 50 organic sessions in the last 12 months.
  • More than 10 referral sessions from reputable sites.
  • Any recorded conversion, lead submission, purchase, demo request, or meaningful assisted conversion.
  • Active paid-media traffic, email links, QR codes, affiliate links, or printed campaign URLs.

These are operating thresholds, not universal SEO laws. A local business may care about 10 lost visits if those visitors were booking high-value services. A large publisher may need a threshold of 500 sessions to make the work economically sensible.

3. Fix internal links before chasing external noise

Broken internal links are fully under your control and often indicate a migration or content-governance problem. A crawl should identify links from key templates, navigation, breadcrumbs, articles, category pages, hreflang references, canonicals, and XML sitemaps.

As a practical rule, fix any 404 linked from pages that receive organic traffic or sit within three clicks of core navigation. Remove deleted URLs from XML sitemaps promptly. Leaving them in a sitemap sends conflicting signals: the site is asking search engines to crawl a URL while the server says it no longer exists.

Soft 404s are often more damaging than ordinary 404s

A soft 404 is not merely a bad error-page design. It is a status-code and content mismatch. Common examples include a “product unavailable” template with almost no useful information that returns 200 OK, a site search page with no results indexed for thousands of queries, or a custom error page served successfully to every invalid URL.

Soft 404s can multiply quickly on ecommerce, marketplace, real-estate, travel, and classified sites. Faceted URLs, expired inventory, empty location pages, and internal search parameters can generate thousands of nearly empty pages. The result is not that every URL will rank badly; it is that crawlers may spend time evaluating low-value URLs while stronger pages receive less attention.

Fix the response before changing the design. If the content is genuinely unavailable and there is no useful substitute, return 404. If the resource was intentionally and permanently removed, 410 is also valid. If a close replacement exists, use a server-side 301 redirect.

Do not assume that a friendly 404 page must return 200 in order to be useful. It can return a real 404 status while still offering navigation, a search box, popular categories, and a route back to the site. Keep the page helpful, but make the status truthful.

The homepage redirect mistake

“Redirect every 404 to the homepage” is popular advice because it appears tidy: visitors avoid an error screen and the 404 report shrinks. It is wrong in most cases.

A homepage rarely matches the intent behind a missing URL. Someone following a link to a specific product, location, article, support document, or comparison page does not receive an equivalent answer by landing on the homepage. They must start their search again, which can increase abandonment and obscure the original problem.

At scale, blanket redirects also make diagnosis harder. A crawler may encounter a mass of unrelated URLs that all resolve to one page. Search engines can treat irrelevant redirects as soft-404-like behavior rather than as useful replacements. The redirect itself is not a repair if the destination is not relevant.

Use a 301 redirect only when the old and new pages satisfy substantially the same intent. Good examples include:

  • An old service URL replaced by a renamed version of the same service.
  • A retired product redirected to its direct successor.
  • A merged article redirected to the consolidated article covering the same topic.
  • A migrated URL structure where the page content remains materially the same.

Popular advice is wrong in another case: redirecting every discontinued product to its category page. This can be useful when shoppers can genuinely choose close alternatives, such as a discontinued running shoe redirecting to the current version or tightly matching range. It is not useful when a specific, unavailable item has no comparable replacement. In that situation, a real 404 or 410 plus a helpful error page is cleaner.

A 30-day process for managing 404 errors

For most established sites, a monthly review is enough. Run it weekly during a platform migration, URL rewrite, large content cleanup, ecommerce catalog change, or redesign. The goal is a controlled queue, not a permanently empty report.

  1. Export errors: Combine Search Console examples, crawl findings, server-log requests, analytics landing pages, and backlink data. Deduplicate by URL path.
  2. Classify each URL: Label it internal-link issue, backlink issue, former traffic page, expired content, bot noise, soft 404, or unknown.
  3. Score priority: Mark URLs high priority when they have a credible backlink, 50-plus organic sessions in 12 months, a conversion, or an active internal link.
  4. Select the correct response: Restore, 301 redirect, correct internal links, return 404, or return 410. Do not default to the homepage.
  5. Test implementation: Confirm that the final URL returns the intended status code. Check both HTTP and HTTPS versions, trailing-slash variants, and query-string behavior where relevant.
  6. Monitor for 30 days: Re-crawl important sections, review redirect chains, and watch whether new 404s originate from templates, sitemaps, campaigns, or external sites.

Set a redirect-chain limit of one hop where possible. If URL A redirects to URL B and B now redirects to URL C, update the rule so A goes directly to C. Chains add latency, complicate reporting, and create unnecessary maintenance work.

When to leave a 404 error alone

Leaving a 404 alone is often the right decision. Do so when the URL was never a legitimate page, has no useful links or traffic, receives only automated probes, has no relevant replacement, or represents content that should stay gone.

This is particularly important after migrations. Teams sometimes create hundreds of speculative redirects to make reports look clean. Those rules require maintenance, can send users to irrelevant pages, and may conceal broken internal linking. A smaller set of well-justified redirects is better than a giant redirect file built from guesswork.

For pages that are deliberately removed because of legal, compliance, privacy, or obsolete-information concerns, do not recreate them solely to preserve a URL. Remove internal references, ensure the intended status code is returned, and document the reason. If a relevant replacement cannot be served safely, the correct response is still a 404 or 410.

Measure success by recovered value, not by zero errors

A healthy site can produce thousands of 404 requests each month without an SEO problem. What matters is whether valuable users, links, and internal crawl paths are hitting dead ends.

Track four measures: the number of internal links pointing to 404s, sessions landing on high-priority 404 URLs, conversions lost or recovered from those URLs, and the number of soft 404 templates returning 200. Review trend lines over 30, 90, and 180 days rather than reacting to a single spike.

A realistic target is not “zero 404 errors.” It is zero unresolved high-value 404s, no accidental soft 404 templates, no broken URLs in sitemaps or major navigation, and no blanket homepage redirects masquerading as a fix. That standard protects search performance while keeping engineering effort focused on problems that actually affect the business.

Related reading

Want the measurement, not the pitch?

Send us your domain. We run the baseline on your category prompts and send back the raw answers alongside the score — you can check our working.

Get an AI Visibility Audit
 +84 34 301 8345

Bạn cần tư vấn chiến lược SEO/AEO/GEO?

Đội ngũ chuyên gia Vidco Group sẵn sàng đồng hành cùng bạn

034.301.8345 Chat Zalo