What is a 404 page, and should I worry about it?
The short answer
A 404 means someone asked for a page that does not exist. A few are normal and harmless — usually mistyped addresses. What matters is whether pages that used to exist now return 404, because that loses the rankings and links they had. Redirect those to their replacements, and make your 404 page helpful rather than a dead end.
On this page
What it means
A 404 is the web's way of saying "there is nothing at this address". Someone followed a link or typed a URL, and the page is not there.
Most sites produce some 404s and always will. Someone mistypes an address, an old link on another site points somewhere that never existed, a bot probes for pages you do not have. Those are harmless and require nothing from you.
The kind that costs you
A page that used to exist and now returns 404.
That page may have had a search position, links from other sites, and bookmarks. All of it is now pointing at nothing. Search engines eventually drop the page, the links stop passing any value, and anyone following an old reference hits a wall.
This happens most often during a redesign, a platform migration, or a tidy-up where somebody deleted pages that "nobody was using" without checking.
The fix: redirect, do not delete
When a page goes away, send its address somewhere sensible with a permanent redirect:
- To its replacement, if one exists — the new version of that service, the updated menu
- To the closest relevant page, if there is no direct replacement
- To the homepage, only as a last resort — it is better than nothing and worse than a relevant page
A permanent redirect passes most of the accumulated value to the new address and means old links keep working. It is a few minutes of setup that preserves years of accumulation.
Finding the ones that matter
Google Search Console reports 404s that Google encountered, which is the most useful list because it reflects pages that were actually indexed.
Your analytics can show 404 page views, which reveals what real visitors are hitting.
Before any migration, export a list of every URL on the current site. That list is what you redirect from. Making it afterwards is guesswork.
Make the 404 page itself useful
Some 404s are unavoidable, so the page a visitor lands on should do something better than apologise:
- Say plainly that the page does not exist, without technical jargon
- Offer the main routes — home, menu, contact, booking
- Include a search box if your site has search
- Keep your normal navigation and branding, so it is obvious they are still on your site
- No humour that gets in the way. A joke is fine; a joke instead of directions is not
The goal is that somebody who hits a dead end still gets where they were going.
What not to do
Redirect everything to the homepage indiscriminately. Search engines treat this as a soft 404 and it frustrates people who wanted something specific.
Leave a deleted page returning 200 with an empty layout. Genuinely worse than a 404 — search engines cannot tell the page is gone and keep it indexed.
Panic about the number. A handful of 404s from mistyped URLs and bot probes is normal on every site. The question is never how many, it is which ones.
The one-minute check
Take the three or four most important pages you have ever had — the ones on printed material, the ones people link to — and request each address.
If any of them 404, that is worth fixing this week. Everything else on this page can wait.
The 404s that matter most, and where they come from
Most broken links on a small business site come from one event: the site was rebuilt and the URLs changed.
That is the case worth planning for, because it is the only one that breaks links at scale. Every bookmark, every link anyone ever shared, every ranking those pages earned — all of it pointing at nothing. The fix is redirects from the old addresses to the new ones, and it has to happen at the move, by whoever is doing the move. Retrofitting it later means first reconstructing a list of URLs that no longer exist anywhere.
So make it a question you ask before a rebuild rather than a problem you find after: which URLs change, and where does each old one send people? A good answer is a list. A vague answer usually means nobody has thought about it, and the cost arrives three months later as a decline nobody can explain.
Helm handles this as part of a migration rather than as an extra — old addresses redirect permanently to their new homes, which is the same reasoning set out in will I lose my Google ranking.