Colophon

How this site is built, and what its own measurements say. The same standard we hold a client's system to: show the evidence, not the adjectives.

Built by hand, on purpose

This site has no framework, no build step, no bundler, and no package manager. It is plain HTML, CSS, and JavaScript, written by hand and committed to a git repository that is the single source of truth for everything you see. To change the site, someone edits a file. There is no compile to trust, no dependency tree to audit, and no node_modules to go stale.

That is a deliberate choice. A firm that tells clients to find out what their system actually does before they modernize should be able to account for every line of its own. Here it can: the repository is the site, and nothing is generated by a tool we cannot read.

Measured, not claimed

Most sites quote a single performance score as if it were a fixed property of the page. It is not. Google's PageSpeed Insights runs in a shared lab with simulated mobile throttling, and on this site's homepage the mobile performance score can swing by dozens of points from one run to the next. So instead of freezing a number this page cannot keep true, here is what actually holds steady, and the live check you can run yourself.

Consistently strong

  • Accessibility — 97-100
  • SEO — 100

Moves with the lab

  • Best Practices — high 70s with the ad tags; 96-100 without
  • Performance — high 40s to mid 90s

Two honest notes on the numbers that move. The first: the performance score is dominated by the test network and by the Google Ads and Microsoft Clarity scripts loading during the measured window -- on a throttled mobile run, that contention sets the number, not the site's own code; desktop, throttled far less, typically lands in the 90s. The second: the Best Practices deduction belongs entirely to those same ad and analytics tags, which set third-party cookies. The default lab still permits third-party cookies, and therefore penalizes them -- that alone drops the score into the high 70s with the tags in place; the site's own code passes the audit clean. Both are business choices made with open eyes, not defects we missed.

The honest way to show a moving number is to point at the meter, not a photograph of it. Run the live check through PageSpeed Insights -- the same public tool anyone auditing the site would use. The accessibility and SEO results hold; the performance figure will be whatever the lab and the network give it that minute.

Progressive enhancement, top to bottom

The baseline is readable and usable with no JavaScript at all. Nothing here needs script to convey its meaning; script only makes it better. Where the platform offers something modern, the site uses it and falls back cleanly where it does not.

  • Speculation-rules prefetching, so the next page's HTML is often already fetched before you click.
  • An AVIF and WebP image pipeline with responsive widths, generated by a hand-run script and committed alongside the original files.
  • A command palette built on the native <dialog> element, so the browser provides the focus handling and the keyboard behavior instead of a hand-rolled imitation.
  • A hash-based Content Security Policy (currently in report-only mode), WCAG AA color contrast, and motion that respects your reduced-motion setting.

The build that runs by hand

There is no build server, but there is discipline in its place. A handful of small Node scripts generate the parts of the site that must not drift -- the navigation, the glossary and its structured data, and the Content Security Policy hashes -- each from a single source. They are run by hand, their output is committed, and a check runs in continuous integration that fails the build if any generated file falls out of sync with its source. The result is a static site with a generator's guarantees and none of a framework's weight.

Why build it this way

A backup that has never been restored is a theory with a timestamp. A performance number that has never been measured is the same kind of claim. This page exists so the claims have receipts. If you would rather see how the work is done than read that it is done well, that is the whole idea -- here, and in the report we deliver.

See the methodology, read the evidence of method, or inspect the SocialGoal sample Modernization Risk Report.