The canonical tag that hid a whole site

The worst faults are the silent ones. Nothing breaks. Nothing looks wrong. The site simply does not appear, and every dashboard tells you everything is fine.

I had one of these on a product of my own. Content going out steadily. Pages loading fast, rendering properly, marked up correctly. Search traffic approximately zero for months.

The cause was a canonical tag pointing the wrong way.

For anyone who has not met one: a canonical tag is a line in the head of a page that tells search engines "this is the real address of this content." It exists so that a page reachable at several URLs does not compete with itself.

Get it backward and you are telling search engines that your page is a duplicate of some other page. So it gets crawled, understood, and filed as a copy. Nothing errors. The page is fine. It is simply disqualified, by its own instruction.

What made it hard to find is that everything visible was healthy. The pages existed. They were indexed in the technical sense. The dashboard reported no issues, because from the outside there was no issue, only a page voluntarily standing aside.

I found it by reading the rendered HTML of a page that should have been ranking and was not. Twenty seconds, once I stopped looking at dashboards.

Two lessons, and the second is the real one.

First: whenever a site is inexplicably invisible, view source before you open another tool. Canonical, robots meta, noindex left over from staging. Those three account for most mystery cases, and all three are invisible from the front end.

Second: dashboards report the errors they were built to detect. A page that is politely excluding itself is not an error. The tools that tell you about problems are the tools that will not tell you about this one. At some point somebody has to read the actual page.