Background

How the experience stacked up

Last updated

  • 2011 — college

    Hand-written front-end, the hard way

    I built HTML, CSS, JavaScript and jQuery without frameworks, because there weren't many worth using yet. Layouts meant floats and clearfixes; responsive design meant arguing with Internet Explorer. I also spent time inside WordPress and Blogger, which is where my first real questions about hosting and page speed came from.

  • 2016 — search

    Eight years of technical SEO that had to work

    I took on search visibility for family business websites, which meant results were not optional and there was no agency to blame if a ranking dropped. On-page structure, site architecture, crawl and indexation problems, and the slow work of holding page-one positions through repeated algorithm updates rather than winning them once.

  • 2019 — production systems

    Java, Python and the cost of running things

    I moved into backend and cloud work: JVM services, Python tooling, and AWS infrastructure at a scale where bandwidth and instance sizing show up on a real invoice. That's where my habit of modelling cost before choosing infrastructure came from — and where most hosting advice I read online started looking naive by comparison.

  • 2022 — applied AI

    Building with models, and watching search change

    I've been working with modern model APIs, and tracking what answer engines do to search visibility. The practical question — how does a page get cited rather than merely ranked — is genuinely unsettled, and most of what I read about it is guesswork dressed up as method.

Why this site exists

Hosting advice is written by marketers who have never deployed an application, and deployment guides are written by engineers who have never watched a ranking drop. This site covers the chain that connects them: infrastructure decides performance, performance decides visibility, and visibility decides whether the work mattered.

Nearly every “best web hosting” article online is an affiliate list assembled from pricing pages. The numbers are copied, the rankings are ordered by commission, and nobody deployed anything. That is not a gap in quality, it is a gap in method — and it is fixable by simply doing the work and publishing what the work showed. The Django hosting cost breakdown and the Next.js cost comparison are what that habit actually looks like applied.

Editorial standards

These are the rules this site is written under. They exist to be checked against.

  • Every number is measured or modelled from published pricing, and carries the date it was verified.
  • Test conditions are stated — region, sample size, cache state, workload — so results can be disputed.
  • Affiliate relationships are disclosed above the first affiliate link on any page that contains one.
  • No provider gets advance sight of results, and no placement is for sale.
  • Unfavourable findings are published unchanged, including for affiliate partners.
  • Outdated comparisons are re-checked or withdrawn, not quietly left online collecting traffic.
On corrections. If a figure here is wrong, it gets corrected in place with a dated note rather than silently edited. Hosting prices move constantly and benchmarks age badly; pretending otherwise would undermine the only thing this site actually offers.

What I work with