This site had reached the point where the content was stronger than the infrastructure around it.

The writing was there. The post archive was there. The Hugo setup was working. But a lot of the details that make a site easier to crawl, easier to navigate, and easier to live with day to day were still behind where they should have been. So I spent some time tightening the foundation.

What I changed

The biggest work fell into a few buckets.

First, I cleaned up the SEO plumbing. The site now has better canonical metadata, Open Graph and Twitter card support, structured data for posts, breadcrumb markup, a dedicated About page for author identity, stronger archive summaries, related-post links, and a lightweight tag taxonomy that is useful enough to act as real topic organization instead of just decorative metadata.

Second, I fixed the image story. Post images are now going through a Hugo-based responsive pipeline so cards and article pages are not serving oversized assets by default. I also did a one-time pass on some of the more static images to reduce obvious waste without turning the site into an asset-management science project.

Third, I added a real theme model. The site now supports light mode, dark mode, and system mode, with the preference stored in the browser and applied before first paint. That sounds small, but it makes the site feel much more current and much less like it is frozen in a single visual assumption.

Finally, I cleaned up some of the content structure itself. The post sidebar is now chronological, the archive cards use tighter summaries, the About page is a real page rather than only a home-section anchor, and Google Analytics is configured in a production-only way instead of being half-configured and half-forgotten.

Why this work mattered

Most of these changes are not glamorous.

They are the kind of things that sit underneath a site and quietly determine whether the content gets presented well, understood well, and revisited later. Search engines care about structure. Social platforms care about metadata. Human readers care about clarity, load speed, readable cards, and whether a site respects the theme setting of the device they are already using.

For a personal site, those details matter even more because there is no product team hiding behind the presentation. The site itself is part of the work.

The dark mode change was worth it

One of the more visible improvements was adding light, dark, and system theme support without forking the upstream Toha theme.

That mattered to me for two reasons. The first is practical: plenty of people browse in dark mode now, and a site that ignores that feels dated. The second is maintenance: I wanted the change to live in repo-owned overrides instead of turning the theme submodule into something I would have to manually untangle later. The result is cleaner than what was there before and easier to maintain than a fork would have been.

The SEO work is more important than it looks

The SEO improvements are not about chasing gimmicks.

They are about making the site easier for Google and other systems to understand. That means better metadata, cleaner summaries, clearer topical relationships between posts, better author identity, and fewer ambiguous signals. I would rather have a smaller set of solid technical signals than a larger set of shallow tricks.

The same goes for the tag taxonomy and related posts. Those changes are partly for readers, but they are also for structure. If I write repeatedly about topics like .NET, Azure, integration, cloud architecture, and AI, the site should reflect that as actual connected subject matter rather than as a pile of disconnected pages.

What I want to see next

I am not going to pretend this work is automatically a success just because it is now implemented.

The interesting question is whether it actually changes anything. Do pages get indexed more cleanly? Do impressions improve? Do more visitors move from one post to another? Do the updated archive summaries and metadata improve click-through? Does a cleaner structure turn into more page views over time?

That is the real test.

I have Google Search Console set up now, and the analytics side is finally in a state where it can tell me something useful. If these changes materially improve traffic or engagement, I will write a follow-up post once there is enough data to say something real instead of just guessing.

For now, this is the groundwork. And frankly, it was overdue.