Skip to content
Performance

Core Web Vitals for WordPress: which decision moves which metric

LCP, INP and CLS translated into the specific WordPress decisions that move them — and the popular fixes that move nothing.

Which WordPress decisions actually move each Core Web Vital?

Mostafa Nouh — portraitMostafa NouhPublished 8 min read

Core Web Vitals get treated as a single score to be raised. They are three unrelated measurements of three unrelated failures, and a change that helps one can hurt another — deferring all JavaScript improves INP and can destroy LCP if the hero image is lazy-loaded behind it.

LCP — the server, then the image

Largest Contentful Paint on a WordPress page is usually two numbers added together: time to first byte, and the delay before the browser discovers the hero image. Fix them in that order. A 900ms TTFB caps your LCP at 900ms no matter what you do to the image.

  • Page caching that actually serves the cached copy — verify with a response header, not the plugin's dashboard.
  • Object caching for the queries that survive page caching, on any store or logged-in area.
  • The hero image in the HTML with fetchpriority="high" and NO loading="lazy".
  • Modern image formats and correct srcset, so the phone downloads a phone-sized file.
  • Self-hosted or preconnected fonts — a font request on the critical path is a real delay.

INP — how much code runs on a tap

Interaction to Next Paint measures the delay between a user tapping and the screen changing. On WordPress the usual cause is quantity: fifteen plugins each attaching their own listeners, a slider library, an analytics stack, and a page builder's runtime — all competing for one main thread on a mid-range Android.

Measure which script owns the long task before removing anything. Deferring a script that was not on the critical path buys nothing, and deleting a plugin whose 4KB of JavaScript was never the problem is how a site loses a feature and keeps its score.

CLS — space that was not reserved

  • Every image and video carries width and height attributes, or an aspect-ratio.
  • Ad slots, embeds and cookie banners get a reserved box before they arrive.
  • Webfonts declare a fallback with matched metrics, so the swap does not reflow the paragraph.
  • Nothing is injected above existing content after load — a promo bar belongs in the HTML.

The lab score is not the metric

Sources

Contact

Have a site or store that needs to work better?

Tell me what is happening and on which page. A diagnosis is a useful first deliverable on its own.

I read every message myself and reply within one working day — usually with a question or two before anything else, because a scope worth quoting needs them answered.

Available for remote work aligned to EST hours
Mostafa Nouh — portrait

Mostafa Nouh

Full-Stack Developer & Product Builder

Location
Egypt · Remote