Editorial · Core Web Vitals

Core Web Vitals Monitoring 2026

The three Core Web Vitals, Largest Contentful Paint (LCP), Interaction to Next Paint (INP), Cumulative Layout Shift (CLS), have unchanged 'good' thresholds in 2026. Every RUM tool captures them. Cost depends entirely on what else you need alongside.

Direct answer

What are the 2026 Core Web Vitals thresholds?

LCP under 2.5 seconds is 'good', 2.5-4.0s is 'needs improvement', above 4.0s is 'poor'. INP under 200ms is 'good', 200-500ms is 'needs improvement', above 500ms is 'poor'. CLS under 0.1 is 'good', 0.1-0.25 is 'needs improvement', above 0.25 is 'poor'. These thresholds have been unchanged since INP went stable in March 2024.

The three metrics defined

LCP (Largest Contentful Paint) — time from page load start until the largest element above the fold is rendered. Typically a hero image, headline, or video poster. Measures perceived load speed.
INP (Interaction to Next Paint) — time from a user interaction (click, tap, keypress) until the next visual update. Measures responsiveness. Replaced FID (First Input Delay) as a Core Web Vital in March 2024.
CLS (Cumulative Layout Shift) — sum of layout shifts during page lifetime, weighted by impact. Measures visual stability. A button that moves down 200px just as you're about to click it is a CLS event.

Tools that capture all three

Every vendor in our matrix captures LCP, INP, and CLS. The differentiator is how:

Cloudflare Web Analytics — free, captures all three, dashboards them but doesn't alert.
SpeedCurve — captures all three plus the broader RUM dataset; performance-budgets feature lets you alert on regression.
Datadog RUM, Dynatrace RUM, New Relic Browser, Sentry, Raygun, Atatus, Sematext Experience, AWS CloudWatch RUM — all capture and alert on Core Web Vitals.
LogRocket, Fullstory — capture them but lead with replay UX rather than CWV-first dashboards.
Grafana Faro — open-source capture; you build the alerting yourself.

The 75th-percentile rule

Google evaluates Core Web Vitals at the 75th percentile of real user experiences. A page passes “good LCP” if 75% of users experience LCP under 2.5 seconds. This matters for tool selection: synthetic monitoring measures a single scripted scenario, RUM measures the actual user distribution. Only RUM data tells you whether 75% of users are within threshold.

INP is the newest and hardest to optimise

INP became a Core Web Vital in March 2024, replacing FID. Where FID only measured the first interaction's latency, INP measures all interactions and reports the worst (or 98th percentile). This is much stricter, pages that passed FID often fail INP, particularly heavy React/Angular applications with synchronous render work blocking the main thread.

For INP-specific monitoring detail, see INP monitoring cost.

What Google uses CWV for

Page Experience signal in Search ranking. Pages passing CWV thresholds get a small ranking boost. Pages failing get penalised. The signal is real but not dominant.
Page Experience report in Search Console. CWV-based site health reports.
Chrome User Experience Report (CrUX). Public aggregated CWV data per origin, used by many tools (PageSpeed Insights, web.dev tester).

When CWV is the only thing you need

For sites that only care about Core Web Vitals (e.g., for SEO signal validation), Cloudflare Web Analytics is free and sufficient. For sites that want CWV plus errors plus replay, you need a commercial RUM vendor. The cheapest viable commercial option for CWV + errors is Raygun Basic at $80/year.

Last verified June 2026