What to do with 404 pages besides apologising for them

Aug 29
3 min read

Updated: 3 days ago

Introduction

An audit reports two hundred error pages and the business assumes something is badly broken. Somebody proposes redirecting all of them to the home page to make the number go away.

Most of those addresses are mistyped links, old test pages and requests from automated scanners that no human will ever follow. A handful are genuinely valuable — pages that still have links pointing at them and people arriving. Separating the two is the whole task, and the blanket redirect is actively worse than doing nothing.

1. What to do with 404 pages starts with who is reaching them

Triage before action.

An error nobody encounters costs nothing. An error reached fifty times a month by people following a link from another website is a real loss, and the two require completely different responses.

2. Find out where the requests come from

The data you need to decide.

Your server logs, analytics and search console each report addresses returning errors, and the better sources also report the referring page. That referrer is what tells you whether a human sent somebody there.

3. Redirect the ones with links or traffic

The valuable minority.

If another site links to a missing page, or people still arrive at it, redirect it to the closest relevant page. This recovers both the visitors and whatever authority the link was passing.

4. Leave genuine errors as errors

A correct response, not a failure.

An address that never existed, or content deliberately removed with no equivalent, should return an error. That is the accurate answer, and it lets search engines drop the address rather than maintaining it indefinitely.

5. Never redirect everything to the home page

The tempting shortcut that backfires.

It removes the number from the report while sending people somewhere that does not answer their request, and search engines treat these as errors regardless. A relevant destination or an honest error is better than either.

6. Make the error page useful

Where the design effort belongs.

A search box, links to the main sections, and a plain explanation recovers a share of the people who land there. A page saying only that something went wrong sends everybody straight back to the results.

7. Check that it returns the correct response

A subtle and common misconfiguration.

Some sites display an error message while technically reporting success, which means search engines keep the address indexed forever. The page must actually return an error status, not merely look like one.

8. Fix the internal links that cause them

The category entirely within your control.

Errors reached from your own pages are links you created and can repair. These should be cleared completely, since they are both a visitor problem and a wasted signal within your own site.

9. Watch for a sudden increase

The early warning that matters.

A gradual trickle is normal on any site. A sharp rise usually means a deployment broke a section or a bulk edit changed addresses, and catching it in the same week is far cheaper than finding it in next quarter's audit.

Keep an eye on the referrers over time. External sites occasionally link to a page you never had, usually because somebody guessed an address or a print advert contained a typographical error, and creating a redirect for a frequently requested one recovers traffic at no cost.

Conclusion

Sort errors by whether anybody is actually reaching them, because that decides the response.

Use logs, analytics and search console to find the referrers, redirect only the addresses with real links or traffic, leave genuine errors returning errors so they can be dropped, refuse the blanket redirect to the home page, make the error page itself helpful with search and navigation, confirm the page returns a real error status rather than reporting success, clear every internal link causing an error, and investigate any sudden increase immediately.