HTTP 404 Bad Request: What the Error Means

Updated September 2026 · Written and maintained by the Progression Agency strategy team

An HTTP 404 bad request is not one error, it is two — and knowing which one you are actually looking at is most of the fix. A 404 means the server understood your request perfectly and has nothing at that address. A 400 Bad Request means the server could not understand the request at all. They have different causes, different fixes and different consequences, and browsers sometimes present them in ways that blur the two. This page separates them, then covers 404 bad gateway confusion, and finishes with what a site owner should actually do about broken URLs.

What is a 404 error?

HTTP 404 Not Found means the server received and understood your request and there is no resource at that address. The request was valid; the address was not. That distinction is what separates it from every other error on this page.

A 404 does not mean the site is down. It means one specific address on it does not resolve. The fastest confirmation is to delete everything after the domain and load the homepage: if that works, the site is fine and the path is wrong.

Common causes of a 404

• A typo or a truncated paste in the address.
• A page that was renamed or moved without a redirect being put in place.
• A page deliberately deleted by the site owner.
• An internal link on the site still pointing at an old address.
• An external site linking to a URL that changed years ago.
• Case sensitivity: some servers treat /Page and /page as different addresses.
• Trailing slash inconsistency between the link and the server’s expectation.

Is a 404 always a problem?

No. Returning 404 for something that genuinely does not exist is correct behavior, and a site with no 404s at all is usually one that redirects everything to its homepage, which is worse. A 404 becomes a problem when it replaces a page that had traffic or inbound links.

What is a 400 Bad Request?

HTTP 400 Bad Request means the server could not understand the request. Something in what your browser sent was malformed — the URL syntax, a header, or most often a cookie that has grown too large. Unlike a 404, this one is frequently caused on your side and frequently fixable in under a minute.

The scorecard above is a triage tool. Anything marked yes is usually resolvable from the browser; anything marked no requires the server operator. The partial rows are the ones worth testing in a private window first, because that single step tells you which side the problem sits on.

Causes of a 400 error you can fix yourself

1. An oversized or corrupted cookie for that specific site. This is the most common cause.
2. A malformed URL, often from a paste that included a stray character or a line break.
3. Invalid characters in the address that were not encoded properly.
4. A request header made too large by a browser extension.
5. A stale cached redirect pointing at a malformed address.

How to fix a 400 error in three steps

1. Open the same URL in a private or incognito window. If it loads, the problem is local state.
2. If it loaded privately, clear the cookies and site data for that one site only.
3. Reload the original window. If it still fails, retype the URL by hand rather than pasting.

Clearing cookies for one site is deliberate advice: clearing everything logs you out of everywhere for no benefit, since the malformed cookie belongs to the site returning the error.

What is the difference between 404 and 400?

A 404 means the address does not exist; a 400 means the request could not be understood. The 404 is about what you asked for, the 400 is about how you asked. That is why the fixes are completely different: one is a URL problem and the other is usually a cookie problem.

The important reading of that table: only the first and fourth rows are frequently fixable by the person seeing them. Everything from 500 downward is a server-side condition, and no amount of clearing cookies will change it.

The chart makes the same point visually. The 400 row scores highest on ‘fixable by you’ and lowest on SEO impact, which is why a 400 is annoying rather than serious. The 404 row is the reverse: harder to fix from the browser, and materially worse for a site if it replaces something that ranked.

What is a 404 bad gateway?

It is a conflation of two separate errors, and if you saw that phrase you almost certainly received one or the other rather than both. A 404 Not Found means the page does not exist. A 502 Bad Gateway means a server acting as a gateway received an invalid response from an upstream server. They come from different layers and have different fixes.

Check the actual status by opening the browser’s developer tools, selecting the network tab, and reloading. The status column shows the real code. This matters because a 502 usually resolves itself and a 404 usually does not.

If it really is a 502

Wait a minute and retry, because 502s are frequently transient. If it persists across several minutes and several pages, the site has an infrastructure problem and the only useful action is reporting it. Nothing on your device will change the outcome.

If it is a 404 after all

Then follow the 404 steps: check the URL, load the domain alone, search the site for the page, and report the broken link to the owner with the exact address and where you found it.

The sequence above is ordered by cost rather than by likelihood, which is why it starts with re-reading the address. The first three steps take under a minute between them and resolve the large majority of cases without anybody else being involved.

For site owners: what to do about your own 404s

Redirect what had value, return 410 for what was retired deliberately, and repair the internal links that pointed at the dead address. Redirecting everything to the homepage is the common shortcut and it is worse than leaving the 404 in place.

The quadrant chart sets the priority. Old pages that carried traffic and backlinks sit top right — high value, easy fix — and should be redirected first. Genuinely retired pages sit bottom left and should return 410 rather than be pointed at something irrelevant.

What that ordering shows is that six of the seven common causes are inside your own control, and the top two are found by crawling your own site rather than by waiting for a visitor to complain. A crawl is the highest-return hour available to most site owners.

Why blanket redirects to the homepage are wrong

Because a redirect from a specific product page to a homepage is, from a search engine’s point of view, a soft 404: the destination does not fulfill the request. It also frustrates the visitor, who now has to search for what they wanted. Redirect to the closest genuine equivalent or return 404 or 410 honestly.

Use the top three rows as a work queue and ignore the bottom one entirely. Sites accumulate 404s from automated scanning that never existed and never mattered, and treating those as work wastes the hour that should go to the first three rows.

When to use 410 instead of 404

When you removed the page deliberately and nothing replaces it. A 410 Gone states that the removal was intentional and permanent, which is a clearer signal than 404 and typically leads to faster de-indexing. For a discontinued product with no successor, 410 is usually the honest answer.

Fix the link, not just the symptom

A redirect makes the error disappear while the broken internal link stays in the page. The next crawl finds it again, and over years a site accumulates chains of redirects that slow everything down. Repair the link in the source page as well as adding the redirect.

Does a 404 hurt your search rankings?

Only when it replaces a page that had traffic or inbound links. A 404 for something that never existed costs nothing. A 404 on a page that ranked and had links wastes both, and that is the case worth fixing quickly.

The last row is the one people get wrong most often. A page that says ‘not found’ while returning a 200 status tells search engines the content is real, which is worse than an honest 404 because it invites indexing of an empty page.

The pattern in that table is that the right answer is almost always the one that describes reality most accurately. Search engines and visitors both behave better when the status code is honest than when it is optimized.

Custom 404 pages are for people, not for search engines

A good 404 page explains what happened in plain language, offers site search, and links to the main sections. It does not change the status code and it should not pretend the page exists. Its job is to keep a visitor on the site rather than to influence anything technical.

How to prevent 404s in the first place

Decide URL structure once, redirect whenever anything moves, crawl regularly, and treat discontinued content as a decision rather than a deletion. Most 404s on established sites come from a migration where the redirect map was an afterthought.

1. Choose a URL pattern at the start and do not change it for cosmetic reasons.
2. Whenever a page moves, create the 301 redirect in the same change, not later.
3. Before any replatform, export every URL with its traffic and map every one.
4. Crawl the site monthly and fix internal links pointing at dead addresses.
5. Check Search Console coverage reports rather than waiting for complaints.
6. Have a written rule for what happens to discontinued products and expired pages.
7. Keep redirect chains short: point old URLs at the final destination, not at another redirect.

Rule seven quietly matters. Chains of redirects accumulate over years, slow every request through them, and occasionally break entirely when an intermediate step is removed. Repointing old URLs directly at the current destination is maintenance worth doing annually.

If you are planning a site move, our web design and development page and ecommerce development page both cover the migration planning that prevents this category of problem entirely.

404 status, and what it means when search returns one

The number itself is a plain HTTP status code, and the confusion is usually about which of several similar situations produced it.

A 404 status means the server understood the request and has nothing at that address. Search 404 results — a 404 search returning the error page rather than results — usually means the site’s internal search is routing to a missing template rather than that nothing matched. That distinction matters: a genuine no-results page should return 200 with a helpful message, because it exists and is working. Returning a 404 status for an empty search result is a configuration error that costs you the chance to keep the visitor.

Error pages on your own site: a five-minute check

Type a deliberately wrong address on your own domain and look at three things: what the page says, whether it offers a route onward, and what status code it returns. Most sites fail at least one of the three, and all three are cheap to fix.

The first row is the one that matters technically and the third is the one that matters commercially. A helpful error page keeps a visitor who would otherwise leave, and an honest status code keeps search engines from indexing an empty page.

Web platform, performance and search talks from the platform publishers

Publicly available sessions on the web platform, performance and search. None of these are ours; each is credited to its channel and upload date, every identifier was checked live before publication, and each tile loads its player only when clicked.

Distinguishing the 400 family from the 404

A 404 says the resource does not exist; a 400 says the request itself was malformed. Both appear to a visitor as a broken page and they have different causes: a 404 is usually a link or a rename, a 400 is usually an oversized header, a malformed cookie or an encoding fault, and it is frequently client-side rather than a fault of the site.

Frequently asked questions