Do 404s Hurt SEO? Read Crawl Stats Before You Panic

Do 404s hurt SEO is the question people ask after Crawl Stats turns red. A pile of Not found rows looks like a penalty. Most of them are the server telling the truth.

This page is Status-Then-Triage. It is not 404 page UX (custom hospitality). It is not broken links in SEO (find and fix the crawler’s path). It is not crawl budget on small blogs (you probably do not have one). It is not why Google isn’t indexing your pages (the full ladder).

Genuine 404s usually do not tank a site. Google says it is not possible, worthwhile, or desirable to fix every missing URL. Read Crawl Stats and Page indexing. Fix 404s you still link or list in a sitemap. Soft 404s and 5xx responses are the real fires.

Official: 404 (Page Not Found) errors, Crawl Stats report, Page indexing, and HTTP status codes.

Table of contents

1. Status-Then-Triage
2. Google’s honest line on 404s
3. Crawl Stats columns that matter
4. A robots.txt 404 is a success
5. Soft 404s and 5xx
6. When a 404 is still your job
7. FAQ
8. Close the report, fix the few

Status-Then-Triage

Original split for this post. Read the status before you rewrite the site.

        STATUS-THEN-TRIAGE
  1. STATUS   → real 404/410, soft 404, 5xx, or 429?
  2. SHOULD   → should this URL exist?
  3. LINKED   → still in nav, posts, or the XML sitemap?
  4. ACT      → ignore / 301 to a replacement / restore / fix the server
  Never       → “zero 404s” as a ranking project

Crawl Stats is Google’s crawling history and how your server answered. It is not a scoreboard. A healthy blog can show 404s every week. Typos, retired slugs, and fantasy URLs from old sitemaps show up because Googlebot tried them.

Google’s honest line on 404s

Google’s 404 help is blunt: 404s generally will not impact search performance. Ignore them if you are sure the URLs should not exist. Return a proper 404. Do not fake a page, dump every miss onto the homepage, or block 404s in robots.txt—those create soft 404s.

If you deleted a post and have no replacement, 404 or 410 is the right answer. Google currently treats 410 the same as 404. The report should drop a gone URL after a while. You do not need a mascot page to “save” a URL you meant to kill.

Google’s crawler status-code docs add the crawl-frequency piece: most 4xx responses (except 429) are treated as “this content does not exist,” and crawl frequency on those URLs decreases. That is expected. 429 is overload—slow down, do not celebrate it as a tidy 404.

Page indexing is allowed to skip a URL for good reasons, including a 404 on a removed page. The job is not “index everything that ever existed.”

Crawl Stats columns that matter

Open Search Console → Settings → Crawl Stats (or the Crawl Stats card, depending on the UI). You are looking at host-level history: requests, response codes, file types, and whether Google could fetch robots.txt.

Spend time on:

• 5xx — the server failed. That can stall or thin out crawl. Pair with uptime, not with a custom 404 redesign.
• 429 — you told Googlebot you were overloaded. Fix rate limits or hosting, then re-check.
• Robots.txt fetch — success vs failure. Failure can slow or stop crawl. A 404 on robots.txt is not a failure.
• 4xx volume — context, not a panic number. A spike after a redesign is a clue to check inlinks and the sitemap. A steady trickle of junk URLs is normal.

Do not manage “crawl budget” from this chart on a small blog. That owner already told you to skip the large-site guide. Crawl Stats here is what the server said, not a request for more Googlebot minutes.

If one URL still matters, leave the chart and use URL Inspection. Charts do not replace a live fetch.

A robots.txt 404 is a success

This line surprises people.

Crawl Stats treats a 404 on /robots.txt as success: there is no robots file, so Google may crawl. Many small blogs never add a robots.txt and are fine. The robots.txt owner is about writing a file on purpose—not about manufacturing one to clear a 404 count.

What is not success: robots.txt returning 5xx, timing out, or oscillating. Google’s Crawl Stats help notes that fetch problems there can slow or stop crawling. Fix the host or the file’s availability. Do not confuse that with a missing blog post.

Soft 404s and 5xx

These are the rows that deserve a calendar block.

Soft 404: Google thinks the URL is empty or “not found” even though the status was not a clean 404/410. Common causes: a 200 template that says “oops,” or a 301 of every typo to /. Google’s 404 help says those patterns make it harder to understand the site. Hospitality on a real 404 template is the UX owner. Pretending the homepage is every missing slug is this problem.

5xx: the page might exist. The server did not prove it. Indexed URLs that flap 500 can drop or get recrawled slowly. That is not a 404 debate.

Crawled – currently not indexed is a quality or timing status, not a 404 count. Stay on that owner if that is the label you actually have.

When a 404 is still your job

Google still wants you to fix 404s you link to or submit in a sitemap. Page indexing help is explicit: gone on purpose is fine; listed as live and returning 404 is a mess you created.

Triage only those:

If a replacement exists, 301 to it—that is the broken-links workflow. If you only need a kinder dead page, use the 404 template. If the whole host moved, that is a domain-move packet, not a Crawl Stats clean-up.

FAQ

Do 404s hurt SEO?

Genuine 404s usually do not tank a site. Google says they generally will not impact search performance. Fix 404s you still link or sitemap. Soft 404s and 5xx are the real fires.

Should I try to fix every 404 in Crawl Stats?

No. It is not possible, worthwhile, or desirable to fix all 404s. Many are correct. Return a proper 404 or 410 and move on.

Is a 404 on robots.txt a problem?

No. In Crawl Stats that 404 is success: no file, crawl allowed. Fetch failures on robots.txt are the problem.

Is this the same as 404 page UX?

No. That page is the template. This page is whether the counts should scare you.

What is a soft 404?

A gone-looking URL that did not return a clean 404 or 410—often 200 “oops” or a homepage dump. Do not fake content or robots.txt-block 404s.

Do 410s rank differently than 404s?

Google currently treats them the same. Use 410 when you mean retired. Not a ranking trick.

Which report should I open first?

Crawl Stats for crawler history. Page indexing for Not found vs soft 404. Inspection for one URL you still care about.

Where is Google’s official guidance?

404 errors, Crawl Stats, Page indexing, HTTP status codes.

Close the report, fix the few

Open Crawl Stats. Separate 5xx and robots fetch failures from ordinary 404s. Then open Page indexing and only chase URLs you still recommend. Leave the rest as honest 404s. A red count is not a ranking verdict—and zero 404s is not a goal Google asked you to hit.