Deeka's Devlog

Deekas

3 min read

Improve SEO metadata, crawlability, and image performance

Summary

Audited the Hugo/PaperMod site using repository inspection, generated HTML, and live crawlability checks, then committed the improvements as 06ac871 on the new seo-improvements branch. Improved the homepage search title without changing article titles. Added a concise CORS description, then used four non-overlapping agents to review all 43 existing content pages: the subsequent batch added 28 descriptions, revised four, and retained 11. Verified unique explicit descriptions, 129 SEO/social metadata checks, and 37 article schema checks; the description batch preserved article bodies and other front matter.

Curated seven indexable topic archives (Java, DevOps, Git, MongoDB, Networking, Security, and Web Development) with descriptions; applied noindex, follow and sitemap exclusion to 79 other archive pages while retaining navigation and article indexing. Replaced homepage Organization structured data with a shared Person identity for homepage and article authors/publishers through local template overrides. Corrected heading hierarchy in the Tmux, Java Singleton, and Jackson articles while preserving existing anchors. Enabled Hugo robots.txt generation to expose the sitemap in production and block development crawling. Replaced the parking article’s 3,404,937-byte PNG with a 1600×900 WebP of 163,326 bytes (95.2% smaller), removed the original source image, and added a Cloudflare Pages 301 redirect for its legacy URL.

Added tests/test_archive_seo.py with 15 build-level regression tests covering production/development indexing, archives, sitemap, shared schema identity, headings, robots.txt, optimized image use, and redirect output. The transcript recorded all 15 tests, the final minified production build, and staged whitespace checks passing; schema-specific checks parsed 82 JSON-LD blocks and confirmed unchanged visible body markup across 132 pages at that stage. Repository review confirmed the commit and a clean working tree. Changes were not pushed or deployed, so live redirect and post-deployment SEO behavior remained unverified. Social preview images were deferred at the user’s request; Search Console indexing and Core Web Vitals were not measured. Existing Hugo language-related deprecation warnings remained.

Decisions

  • Kept changes on seo-improvements and committed without pushing or adding a social preview image.
  • Defaulted new taxonomy archives to noindex and sitemap exclusion until explicitly curated; preserved browsability rather than removing tags or articles.
  • Used local PaperMod template overrides and one shared Person identity for consistent homepage, author, and publisher structured data.
  • Removed the unused published PNG while preserving its public URL through a deployment-time 301 redirect.

Learnings

  • PaperMod already supplied the sitemap declaration in its robots template; enabling Hugo robots.txt generation was sufficient.
  • Hugo publishes unused files under static, so switching the article reference alone would not remove the original image from deployment.
  • PaperMod also displays page descriptions beneath titles, so description changes were not exclusively hidden metadata.