CSS Wizardry

Safari 27 adds scroll anchoring, tighter HTTP caching, and static Service Worker routes, while Chrome improves performance tooling.

CrUX adds public ad count, density, CPU, and network metrics, giving publishers and advertisers a shared view of real ad weight.

CrUX shows another fall in good INP experiences, making careful attribution more useful than a quick explanation.

Chrome 153 starts exposing dedicated-worker congestion to RUM, while browser releases tighten privacy, module recovery, and image preloading.

Chrome and Firefox are moving to two-week releases, making browser cadence part of how teams interpret RUM and plan performance testing.

Use Gmail’s Message-ID to point Codex or Claude at one exact email and retrieve the complete conversation rather than a plausible result.

Unattributed Navigation Overhead exposes the part of Time to First Byte that Navigation Timing cannot explain, including hidden cross-origin redirects.

LUX Sidecar adds context to RUM, while browser-tool fixes show why measurement conditions and small debugging details matter.

LUX Sidecar is a small SpeedCurve RUM plugin that adds browser and navigation context to your real-user performance data.

Safari Technology Preview 250 fixed a subtle Web Inspector bug that overwrote page-authored Cache-Control request headers when the resource cache was disabled.

A quiet week in web performance is a useful chance to investigate the evidence and problems teams already have.

Chrome 151 and web-vitals v6 bring native SPA measurement into production, while new browser controls reshape device and third-party governance.

CrUX’s seasonal decline, Safari 27’s user-experience fixes, richer INP attribution, and long-session third-party costs all demand better interpretation.

Measure how entire components render with the Container Timing API: how it differs from Element Timing, where it helps, and how to try it today.

The 2026 update to my real-world device testing recommendations paints a clearer picture than ever…

Chrome 151 brings native soft-navigation measurement closer, alongside better redirect timing, revealing CrUX data, and more trustworthy browser tooling.

The TBT Window is the FCP-to-TTI interval used to calculate Total Blocking Time.

Obs.js is a tiny inline script that helps you adapt your site to real-world network, battery, CPU, and memory conditions.

No-Vary-Search lets HTTP caches ignore irrelevant query parameters such as UTM tags, while still keeping meaningful ones like product variants in the cache key.

A quick but important reminder that font-family declarations don’t inherit fallback stacks the way many developers assume.

CSS containment lets you isolate layout and paint work to self-contained ‘islands’.

If your design system can only apply `loading=lazy` or `fetchpriority=high` blindly, it may be safer not to apply them at all.

Obs.js is a tiny inline script that exposes network, battery, CPU, and memory signals to CSS and JavaScript so you can adapt to user context.

Discover the most representative low- and mid-tier mobile devices for web performance testing in 2025.

How fast are pro cycling teams’ and manufacturers’ websites?

Making sense—and use!

Learn how the Cache-Control request header works, how browsers handle refresh and hard refresh caching, and when developers should use it themselves.

Is HTML a programming language?

What is the real, long-term cost of adopting a JavaScript framework?

I’ve recently decided to apply the permissive MIT License to all content on CSS Wizardry by default.

The new Speculation Rules API is incredibly powerful, but we can do so much more!

Why design another new performance score?!

If, like me, you frequently require the Core Web Vitals colour palete, here it is!

If you want to start consulting, you’re gonna need some paperwork!

We can’t do much to change latency, so how can we work around it?

Quantifying the importance of caching just got a lot easier

Why on earth would you make something render-blocking?!

We’re probably familiar with preconnect, but are we getting it right?

We know we should do it, but do we know how?

How long?!

There’s a super quick and easy way to clear cache on your customers’ devices.

Can Low-Quality Image Placeholders and LCP play nicely together?

There’s still a lot of misunderstanding about CWV for SEO.

If HTTP/2 is so much better, why does it look so similar to HTTP/1?!

Is there any reason to still measure the DOMContentLoaded event?

Revisiting and remapping my Site-Speed Topography technique for assessing web performance at large

We’re often told not to use document.write(), but… why?!

Async snippets used to improve performance, but now they’re a legacy anti-pattern.

Critical CSS promises faster loading, but is it worth the complexity?

When implementing performance fixes, it’s imperative that you measure the right thing—but what is ‘right’?