In June I fixed a real problem on a small editorial site I run.
The site was a client-side app — the kind where the server sends a nearly empty HTML shell and the browser assembles the actual page afterward. That works fine for people. It works badly for search crawlers, because the description a search engine shows underneath your link in the results comes from the HTML the server sent. Every page on this site was sending the same generic one. Fifty-odd pages, one description, all identical.
So I built a step into the deploy that renders each page's real title and description into the file the server hands out. Crawlers started seeing per-page text instead of one repeated boilerplate line.
It worked. I can show you that it worked.
The numbers said yes
