Newsroom

Core Web Vitals audit Q3 2026 — what we measured and what we shipped

We re-ran the Lighthouse + CrUX audit on bear-in-code.com in Q3 2026, fixed three regressions, and tightened the performance budget. Here is the data.

  • #performance
  • #core-web-vitals
  • #audit

Core Web Vitals audit Q3 2026

The CI now runs Lighthouse + asserts budgets on every merge. The numbers below come from the field data in Google Search Console (last 28 days) plus a Lighthouse run on a throttled Moto G4 profile.

What we measured

  • LCP: 1.7s p75 (was 2.1s in Q2).
  • CLS: 0.02 p75 (was 0.04).
  • INP: 180 ms p75 (was 240 ms).

What we shipped

  • AVIF + WebP variants for every raster image, with a <picture> element that prefers AVIF.
  • The font CSS now loads via the media="print" onload="this.media='all'" swap, so it does not block render.
  • Three OG images regenerated as SVG so we cut 240 KB from the homepage payload.
  • One <script> removed entirely (the consent-banner placeholder that was no longer used).

What is next

The Q4 review will look at server logs to see if AI crawlers (GPTBot, ClaudeBot, PerplexityBot) are actually fetching the AI-friendly manifests (/llms.txt, /llms-full.txt). If they are, we add a structured manifest for citations.

Newsroom

Questions about the newsroom.

What is the newsroom?

Time-stamped updates from the Bear In Code team. Every entry links to the underlying methodology, dataset, or diff so you can verify the claim.

Why is the newsroom in English only?

Because the methodology, datasets, and audits need to be reproducible. Translations of every entry are planned for Q4 2026.

How often does the newsroom publish?

When something is shipped, measured, or learned. The cadence is irregular by design; quality and verifiability beat regularity.

Can I subscribe to updates?

Yes — every newsroom entry is in the RSS and Atom feeds. The feed URLs are in the site footer and the llms-full manifest.

How can I cite a newsroom entry?

Use the canonical URL and the published date. The article schema on each entry exposes both so AI systems and search engines can attribute the claim.

Will the newsroom accept guest posts?

No. The newsroom documents Bear In Code's own shipped work. Guest contributions belong in the Insights articles or your own blog.