Optimizing Core Web Vitals

By Michael Samuel Naeem · May 8, 2026 · 7 min read

Search Console flagged 40% of our product pages as "Needs Improvement" on LCP. The hero image loaded from a third-party CDN with no priority hint, the web font blocked rendering for 800ms, and a cookie banner injected above the fold pushed content down after paint. Three targeted fixes — preload, font-display swap, and reserved banner space — moved 85% of URLs to "Good" within two weeks.

The three metrics

Each metric captures a distinct user experience dimension. Optimizing one doesn't automatically fix the others.

Measuring field data

CrUX via PageSpeed Insights:

https://pagespeed.web.dev/analysis?url=https://yoursite.com

Search Console → Experience → Core Web Vitals shows URL groups by status.

Real User Monitoring (RUM):

import { onLCP, onINP, onCLS } from 'web-vitals';

function sendToAnalytics({ name, value, rating }) {
  analytics.track('web_vital', { metric: name, value, rating });
}

onLCP(sendToAnalytics);
onINP(sendToAnalytics);
onCLS(sendToAnalytics);

Deploy RUM to catch regressions before CrUX reflects them (28-day lag).

Fixing LCP

Common LCP elements: hero images, heading text, video posters.

Preload the LCP resource:

<link rel="preload" as="image" href="/hero.webp"
      imagesrcset="/hero-800.webp 800w, /hero-1200.webp 1200w"
      imagesizes="100vw" />

Eliminate render-blocking resources:

<!-- Defer non-critical CSS -->
<link rel="stylesheet" href="/styles.css" media="print" onload="this.media='all'" />

<!-- Async/defer scripts -->
<script src="/analytics.js" defer></script>

Optimize the LCP element directly:

• Serve images in AVIF/WebP with responsive srcset
• Inline critical CSS for above-the-fold content
• Use a CDN with HTTP/2 or HTTP/3
• Reduce server response time (TTFB) — LCP can't start until HTML arrives

Fixing INP

INP replaced First Input Delay (FID) in 2024. It measures all interactions, not just the first.

Break up long tasks:

// Bad: blocks main thread for 200ms
processLargeDataset(data);

// Good: yield to browser between chunks
async function processInChunks(data, chunkSize = 100) {
  for (let i = 0; i < data.length; i += chunkSize) {
    processChunk(data.slice(i, i + chunkSize));
    await new Promise(r => setTimeout(r, 0));
  }
}

Reduce JavaScript execution:

• Code-split and lazy-load non-critical JS
• Defer third-party scripts (chat widgets, analytics)
• Use web workers for heavy computation

Optimize event handlers:

// Debounce expensive handlers
searchInput.addEventListener('input', debounce(handleSearch, 150));

Fixing CLS

Layout shifts happen when elements load or resize without reserved space.

Size images and embeds:

<img src="/photo.webp" width="800" height="600" alt="Product" />

Reserve space for dynamic content:

.ad-slot {
  min-height: 250px; /* matches ad unit size */
}

.cookie-banner {
  height: 60px; /* fixed height, no content push */
}

Avoid inserting content above existing content:

• Don't inject banners or notifications above the viewport fold after load
• Use font-display: swap with size-adjust fallback fonts
• Load web fonts with preload to reduce FOIT/FOUT duration

Prioritized fix checklist

1. Measure field data — identify worst-performing URL groups
2. Fix TTFB — server caching, CDN, edge rendering
3. Preload LCP image — single highest-impact LCP fix
4. Eliminate layout shifts — explicit dimensions on media and ad slots
5. Reduce JS payload — code splitting, defer third parties
6. Break up long tasks — improve INP on interactive pages
7. Monitor continuously — RUM alerts on metric regressions

Attribution debugging

The web-vitals library reports element attribution for LCP and INP:

onINP((metric) => {
  const { interactionTarget, interactionType } = metric.attribution;
  console.log(`Slow ${interactionType} on`, interactionTarget);
});

Ship attribution to your analytics platform to prioritize fixes by impact — which pages, which elements, which interaction types hurt most users.

CrUX vs. Lighthouse gaps

A page scoring 95 in Lighthouse can fail CrUX if real users hit it on slow devices, with extensions installed, or from distant geographies. Always validate fixes against field data after deployment — CrUX updates on a 28-day rolling window.

Setting team SLOs

Example internal SLOs tied to CrUX:

• Marketing origin: LCP p75 < 2.0s, CLS < 0.05, INP < 150ms
• App authenticated shell: INP p75 < 200ms (LCP less SEO-critical)

Error budget: two consecutive weeks failing CrUX "Good" threshold triggers perf sprint before new feature work on affected templates.

Competitive benchmarking

PageSpeed Insights compares origin to similar sites—use for stakeholder communication, not absolute targets. Retail peers with heavier images may have worse LCP; beat your own baseline week-over-week first.

CrUX vs RUM cohort mismatch

Search Console CrUX is 28-day rolling origin-level — your RUM can slice by route today. Do not panic on CrUX lag after fix; verify RUM p75 first. Lab Lighthouse is regression gate only — not SLO.

INP interaction targets

Prioritize CTA buttons, menus, comboboxes — fix accessibility keyboard path and INP together. Soft navigations in SPA need custom INP instrumentation — default web-vitals library may miss client routes.

Practical follow-through (1)

Ship the smallest vertical slice first — one route, one widget, one index configuration — with rollback documented before expanding scope. Baseline the user-visible metric this work protects (latency, recall, conversion, task success rate) for seven days before change and seven days after in your largest market.

Compare canary p75 to control before full rollout. Exercise edge paths manually: refresh, back navigation, double-submit, offline mode, and keyboard-only flows. When assumptions change — traffic doubles, vendor upgrades, org restructure — revisit whether the original design still fits; quiet periods hide drift until the next incident.

Practical follow-through (2)

Ship the smallest vertical slice first — one route, one widget, one index configuration — with rollback documented before expanding scope. Baseline the user-visible metric this work protects (latency, recall, conversion, task success rate) for seven days before change and seven days after in your largest market.

Compare canary p75 to control before full rollout. Exercise edge paths manually: refresh, back navigation, double-submit, offline mode, and keyboard-only flows. When assumptions change — traffic doubles, vendor upgrades, org restructure — revisit whether the original design still fits; quiet periods hide drift until the next incident.

Practical follow-through (3)

Ship the smallest vertical slice first — one route, one widget, one index configuration — with rollback documented before expanding scope. Baseline the user-visible metric this work protects (latency, recall, conversion, task success rate) for seven days before change and seven days after in your largest market.

Compare canary p75 to control before full rollout. Exercise edge paths manually: refresh, back navigation, double-submit, offline mode, and keyboard-only flows. When assumptions change — traffic doubles, vendor upgrades, org restructure — revisit whether the original design still fits; quiet periods hide drift until the next incident.

Practical follow-through (4)

Ship the smallest vertical slice first — one route, one widget, one index configuration — with rollback documented before expanding scope. Baseline the user-visible metric this work protects (latency, recall, conversion, task success rate) for seven days before change and seven days after in your largest market.

Compare canary p75 to control before full rollout. Exercise edge paths manually: refresh, back navigation, double-submit, offline mode, and keyboard-only flows. When assumptions change — traffic doubles, vendor upgrades, org restructure — revisit whether the original design still fits; quiet periods hide drift until the next incident.

Practical follow-through (5)

Ship the smallest vertical slice first — one route, one widget, one index configuration — with rollback documented before expanding scope. Baseline the user-visible metric this work protects (latency, recall, conversion, task success rate) for seven days before change and seven days after in your largest market.

Compare canary p75 to control before full rollout. Exercise edge paths manually: refresh, back navigation, double-submit, offline mode, and keyboard-only flows. When assumptions change — traffic doubles, vendor upgrades, org restructure — revisit whether the original design still fits; quiet periods hide drift until the next incident.

Resources

• web.dev Core Web Vitals
• Chrome User Experience Report
• web-vitals library
• Search Console Core Web Vitals report
• INP optimization guide