Mattress brand — making blog posts and boards visible in search

A homepage built quickly with a design-first approach. The content was there, but search engines almost never surfaced it.

This site was a brand homepage for a mattress company. The design was clean, and the informational posts about mattress types, materials, and care were well written. But the client's frustration was that none of it was showing up in search.

The cause became clear the moment I looked at the structure. To ship the design quickly, body content had been dropped into meaningless div blocks as textboxes. Search engines parse body content through semantic structure, so when everything is rendered as flat divs, there's no way to tell what's a heading and what's body text. To a human reader the posts looked well organised; to a search engine they looked nearly blank.

The board itself had a structural problem too. The board index was rendered dynamically by JavaScript, so the first-response HTML didn't contain the post list for the search engine to see. Going into the individual post detail pages, the title tag was the site-wide title ("Mattress Brand — Official Homepage"). In other words, all 100 posts were exposed to search with the same title, making it impossible to tell which post was about what in the SERP.

One more issue — OG meta and meta descriptions weren't being auto-generated per post, so when a post was shared on social media the preview only showed the generic site-wide image. Even with good content, the sharing layer wasn't working, which kills organic traffic from social.

We set the work priorities this way. First, get titles and meta auto-generated per post. Second, fix the body markup with semantic HTML (h1/h2/article/section). Third, make the board index visible to search engines via SSR or a static fallback. Fourth, auto-include post URLs in the sitemap.

We started with per-post meta auto-generation. The post title flowed straight into the page title ('{post title} | Mattress Brand'), and the meta description was extracted automatically from the first 150 characters of the body. OG meta was switched to prefer the post's first image. That single change alone made each post individually identifiable in search results.

The semantic HTML rewrite took longer. We restructured the existing div-based content into article > h1 > h2 > p, and added Article JSON-LD separately to carry the metadata (publish date, author, category). With this in place, the search engine can clearly recognise the body as "article" content, and date metadata transmits accurately.

The board index page had its JavaScript dependency reduced so the post list ships in the first-response HTML. Search engine bots have execution constraints around JavaScript, and they tend to be especially slow to index new posts via JS-rendered listings, so cleaning this up mattered.

Finally, sitemap.xml was switched to a dynamic generation handler so every newly published post is included automatically. The client doesn't have to submit a separate index request for each new post — it flows naturally.

By the time we wrapped, the directly verifiable changes were unambiguous. Around 100 post pages each had unique titles, meta descriptions, and OG images. The board index was exposing the post list in first-response HTML. The sitemap was auto-including post URLs. We'd also fixed some things outside the original scope, which is why the client's review used the phrase "beyond what I expected."