A 404 is someone asking for a page that isn’t there. Sometimes it’s a reader who clicked a link in one of your own posts and hit a dead end. Far more often it’s a script guessing at addresses, looking for a backup file or a login page from software you’ve never installed.
Both show up as “page not found”, and a raw list of 404s mixes them together. The work is telling them apart, so you can spend ten minutes fixing the links readers actually hit and ignore the rest.
Where the real ones come from
On most WordPress sites, the 404s readers hit come from a few predictable places:
- A post was deleted or unpublished, and other posts still link to it.
- A link in a post has a typo, or points at a draft’s address from before it was published.
- The permalink structure changed, for example from dated addresses to plain ones, and old links elsewhere still use the old form.
- A page moved, say from a post to a page or under a new parent, and its address changed with it.
- Images or files were deleted from the media library while posts still embed or link to them.
Changing a post’s slug is usually fine. WordPress remembers the old slug and redirects it to the new one on its own. The trouble starts when a post is gone altogether, or when its address changes in a way WordPress can’t trace.
Why analytics beats the alternatives
Your server’s access logs record every 404, which is the problem: on a typical site, scanners and crawlers make most of those requests. Search Console’s page indexing report shows the 404s Google found, which matters for search but says nothing about what readers ran into.
An analytics tool that filters bots gets closer to the question you care about: which missing pages did people hit, and what sent them there.
Find them in Vireo
In Vireo Analytics, open the Overview and go to the Issues tab. The tab appears once there’s something on it. The Broken links card lists missing addresses for the date range, and splits them into two groups once it can tell them apart.
The first group is links on your own site that lead to a missing page. Each row shows the Missing page, the page it was Linked from, and the number of Hits. These are the ones to fix first, because the broken link is on a page you control and every hit is a reader already on your site.
The second group is “Requests from elsewhere”: nothing on your site links to these. Each row shows the address, the referring site if there was one, and the hits. Most are automated scanners. A few are stale links on other sites, and those are worth a redirect.
Vireo records a 404 on the server, when WordPress decides there’s no page to show, so it works on every visitor whether their browser runs JavaScript or not. Requests from known bots are skipped, and so are your excluded IP addresses and paths. Scanners often forge the referrer header to look like an internal link, so Vireo ignores a referrer that points at the missing page itself, or at another address it has already logged as missing.
Fix them in this order
Start with the links on your own site. Open the page in the Linked from column, find the link, and either correct it or remove it. If the post it pointed to is gone for good, link to the closest thing you still have.
Next, look at Requests from elsewhere that came from a real referring site. Visit that site and find the link. You can rarely edit it, so add a 301 redirect from the missing address to the page that replaced it. Vireo shows you what’s broken, and a redirect plugin or your server makes the fix. Use a redirect plugin, your host’s redirect tool or your server configuration.
Then scan the requests with no referrer. Most of them look obviously machine-made: /wp-config.php.bak, /.env, admin paths for other platforms. Leave those alone. A redirect for a scanner’s guess achieves nothing, and a page that answers 404 is the right response. The ones worth a second look are addresses that look like your own old content, such as a dated post address from before a permalink change. Those are usually bookmarks or links in old emails, and a redirect gives those readers the page they wanted.
Finally, check the Hits column before you spend time on any single row. A missing page with one hit in a month can wait. One with a steady trickle every day is a link somewhere that keeps sending people.
Check the fix worked
Pick a range that starts after your fix, and the address should stop appearing in the list. The old hits stay in the history for the dates they happened, so a range that includes the days before the fix still shows them.
For a big cleanup, export the list instead of working through it on screen. Under Settings, Import & export, Export CSV downloads Broken links for any date range. If you’ve connected an AI assistant to Vireo, it can read the same list and help you sort it. See asking Claude about your analytics.
Limits worth knowing
To stop a flood of scanner requests filling your database, Vireo records at most 500 new missing addresses a day. Addresses already on the list keep counting after that, and a developer can change the limit with the vireo_404_max_new_rows_per_day filter.
If you imported your history from another stats plugin that kept missing pages, they land in this same report. The import comparison shows which sources bring them across.