DD DevDome Track. Protect. Grow.
Redirects & Links

How to Set Up a 404 Monitor in WordPress and Turn Misses Into Redirects

Your site returns 404 hundreds of times a day. A few of those are readers who wanted a page you moved. The monitor is how you find them.

On this page
  1. What the log should record
  2. Separate the probes from the misses
  3. Turn misses into redirects
  4. What the referrer column tells you
  5. Reading the log on a shop or a large site
  6. Fix the source where you can
  7. The monthly routine

A 404 monitor for WordPress records every request for a page that does not exist, so you can find the ones that matter. WordPress itself says nothing; it shows the not-found template and forgets. Meanwhile readers arrive from old links, bookmarks and mistyped URLs on other sites, hit the error page and leave. The monitor turns that silent loss into a list, and the list into redirects. Here is how to set one up and how to read it without drowning in bot noise.

What the log should record

For each 404: the requested URL, the time, the referrer if any, and whether the requester looks like a bot or a human. The referrer is the clue to where a bad link lives; the bot flag is what makes the log readable. DevDome Link Monitor keeps a live 404 log with exactly that split and offers one-click redirect suggestions for the common misses, alongside its content link scanner. The log needs no account. We make it, so weigh that; a 404 log without a bot split is a log you will stop reading within a week.

Keep the log bounded. A retention window of a few months is plenty; the useful 404s show up repeatedly, and the one-off probes can expire.

Separate the probes from the misses

Open the log after a few days and sort by frequency. Two populations appear immediately.

Bot probes. Requests for wp-login.php variants, xmlrpc.php, .env, wp-config backups, phpinfo, and paths of plugins you never installed. Thousands of them, from datacenter ranges, with no referrer. These are scanners doing their rounds. Ignore them in the 404 sense; if the volume is high enough to matter, bot protection that recognises scanners by behaviour is the fix, not redirects.

Human misses. Requests with a referrer, from residential networks, for URLs that look like pages: an old permalink structure, a post slug you changed, a product you retired, a category you merged, a link someone mistyped in a newsletter. Fewer, and every one is a reader you lost.

With the bot split, the second list is short enough to work through in one sitting.

Turn misses into redirects

For each human 404, decide the destination:

  • The page moved. Redirect to its new URL. Slug changes and permalink structure changes are the bulk of these.
  • The page was merged or replaced. Redirect to the page that now covers the topic.
  • The page was deleted and a category fits. Redirect to the category or the closest relevant post.
  • Nothing fits. Serve a 410 so search engines stop asking, and leave it.

Never redirect to the homepage by default. Search engines treat an irrelevant redirect as a soft 404, and the reader still did not get what they came for.

Add the rules in a redirect manager rather than by hand in .htaccess, so you can see click counts per rule. DevDome Redirect Manager stores unlimited rules for free, with 301, 302, 307 and 308 per rule and automatic cache purging across ten cache plugins so a stale cached 404 page does not keep serving after the rule exists.

What the referrer column tells you

The referrer is the most underused field in a 404 log. A referrer from your own domain means an internal link points at a page you moved; fix the link and the 404 stops at the source. A referrer from a newsletter or a social network means a campaign link was mistyped or the page was renamed after sending; a redirect rescues the clicks that are still coming. A referrer from another site means someone linked to you and the link is now dead; the redirect preserves whatever value that link carries, and if it is a site you care about, a short email asking them to update it is worth the two minutes. No referrer at all, combined with a datacenter origin and a suspicious path, is a bot and goes in the ignore pile.

Reading the log on a shop or a large site

On a WooCommerce site the 404 log fills with retired product URLs, old category slugs and variations of search strings. Treat retired products like any deleted page: redirect to the replacement or the category, and 410 the ones with no equivalent so search engines stop crawling them. On a large content site with thousands of posts, sort the human 404s by count and work only the top fifty each month; the long tail of single hits is rarely worth a rule. In both cases keep the retention window a few months and let the rest expire, so the log stays a list you will actually open.

Fix the source where you can

A 404 with a referrer from your own site is an internal link pointing at a page you moved. Fix the link in the post as well as adding the redirect; readers go straight to the right page and the hop disappears. A referrer from another site is a link you cannot edit; the redirect is the fix, and if it is a valuable link, a polite email asking the site to update it is worth sending.

The monthly routine

New 404s appear after every change: a batch of slug edits, a content prune, a plugin that generated URLs you removed, a newsletter with a typo. Once a month:

  1. Open the 404 log, human hits only, sorted by count.
  2. Add redirects for the new real misses.
  3. Fix any internal links that caused them.
  4. Glance at redirect click counts; a rule with heavy traffic means a link somewhere is worth updating at the source.

Ten minutes, and the readers who arrive from old links land where they meant to.

Disclosure: DevDome publishes this blog and makes the products it mentions. We describe what they do rather than promise results.

Key takeaways

  • WordPress serves 404s silently. Without a log you never learn which old URLs people still request.
  • Most 404s are bots probing for login pages, config files and plugin paths. A monitor that separates bot hits from human hits saves you reading them.
  • The human 404s are the list to act on: old permalinks, deleted pages, mistyped links on other sites.
  • Each real miss gets a redirect to the closest live page, never the homepage, or a 410 if nothing fits.
  • Check the monitor monthly. New 404s appear after every slug change, content prune and plugin swap.

Sources

  1. WordPress.org: Creating an error 404 page : how WordPress handles a not-found request
  2. Google Search Central: HTTP status codes and crawling : how search engines treat 404 and redirects
  3. MDN: 404 Not Found : the definition

Links last checked October 17, 2026.

Frequently asked questions

Does WordPress log 404 errors?

No. It renders the 404 template and moves on. Your web server's access log records the request with a status code, but reading it is tedious and it does not know a bot from a reader. A monitor plugin keeps a readable list inside WordPress.

Why are there so many 404s for pages I never had?

Bots probe every WordPress site for wp-login variants, xmlrpc.php, .env, backup files and known vulnerable plugin paths. They generate most of the 404 volume on most sites. They are noise unless the volume is affecting performance, in which case bot protection is the fix.

Should every 404 get a redirect?

No. Redirect the URLs humans request where a relevant page exists. Leave bot probes alone. For deliberately removed pages with no equivalent, a 410 tells search engines to stop asking.

Will a 404 monitor slow the site?

A well-built one writes a row only when a 404 happens and nothing on normal page views. Keep the log bounded with a retention window so the table does not grow forever.

DevDome Team WordPress plugin builders

The DevDome team builds lightweight, performance-first WordPress plugins and free tools for site owners, founders and marketers. We write about the exact problems our plugins solve, in plain language for people who run real sites.