The made-up address
Once per scan, after the pages are read, SEOmise asks your site for one address no site uses, like /seomise-check-1a2b3c-not-a-page (a new random code each time). It asks only your own site, obeys your robots.txt and follows no redirects.
- It answers 404 (or 410): as it should. Nothing to do.
- It answers 200 OK: Missing pages answer OK. Every mistyped or deleted address looks like a working page to search engines: they keep crawling such addresses and can index pages with no content.
Pages that show your "not found" page
The page your site answered with is compared with the main text of every page of the scan (in short pieces of three words, so menus and footers do not count, and the address shown on the page is left out). A page that answers 200 but shows the same "not found" text is a Soft 404 or broken page. So is a page our browser check found showing an error or nothing (see "What is the browser check in my reports?").
They are grouped by template: the kind of page your site labels it (post, product…) or the first part of its address (/blog/*), since one cause usually explains a whole template. On the site map, open a page to see how sure SEOmise is and how many pages of the same template show it too.
How sure is it?
- A page opened in our browser that shows the "not found" page: 90%.
- Both the "not found" page and our browser check agree: up to 95%.
- A page read only as your server sent it: 75% (a setting), since scripts may change what visitors see.
- A page that rendered only partly: at most 50%.
Below 70% a match says Needs a re-check and costs nothing. If your start page itself matches the "not found" page (an app that answers every address with the same page), the plain read is not used for matching; only pages opened in our browser count.
How to fix it
Each finding has the three usual choices, with steps for your platform:
- WordPress: look for a redirect or "404 to home page" plugin, or a theme whose 404 template answers 200.
- Shopify, Wix, Squarespace, Webflow and other hosted builders: the platform answers 404 itself; check for a custom 404 app or a catch-all redirect.
- Laravel and plain PHP: a catch-all route that shows a page without answering 404 (
abort(404)). - JavaScript apps: the server or host sends every address to
index.html; answer 404 for unknown routes on the server, or add a noindex robots tag on the app's not-found screen. - Any site, through the universal connector or by hand: keep your friendly "not found" page for visitors; only the answer code changes. Redirect a moved page to its new address instead.
The next scan asks for a new made-up address and reads the pages again.
Settings
Platform admins can switch the made-up address off, change its path (on your site only, with {random} once), and set how alike the text must be (85%) and how sure a plain-read match is (75%), under Settings → Policies → Website scan checks.
Not what you needed? Ask Mise or contact support.
