Why a website can be invisible to Google without anyone noticing

A website can work perfectly for its visitors and be completely absent from search engines. This is not about keywords or tags: it is about who builds the page.

The symptom

When a page is assembled by the visitor's browser, the server sends nothing but an empty shell. The visitor sees nothing wrong — their browser does the work. A search engine receives that shell and finds almost nothing to index.

On this very site, before it was reworked, the served HTML contained 104 characters of text for a 407 KB home page. Everything — services, presentation, contact details — existed only after browser-side code had run.

The trap: silent failure

Server-side rendering can be fixed, but it fails badly. When a component throws during server rendering, most frameworks produce no error and no blank page: they silently fall back to browser rendering. The site works, nobody notices, and the problem remains.

We hit two such cases here. The most instructive came down to one character: a component read window?.location. Optional chaining guards against a missing value — it does not guard against a variable that does not exist at all. On the server, window does not exist: the code threw, and the whole site fell back to browser rendering.

The method lesson

Because the failure is silent, you cannot verify it by the absence of errors. You need positive proof: count the characters actually served, look for an expected heading in the raw HTML. "It did not crash" proves nothing.

The outcome

After the fix, this site's home page serves over 1,700 characters of real text to search engines, with its titles, descriptions and structured data — across twelve pages, in French and English.

This is the kind of verification we run on every platform we build: a site nobody can find is not a site, it is a brochure.