First Contentful Paint


On this page
Available in:
SyntheticCrUXRUM

First Contentful Paint (FCP) measures the time from the start of navigation until any of the page's content first appears on screen, such as text or an image. 1.8 seconds or less is good, and above 3.0 seconds is poor.

FCP is a supporting Web Vital, not one of the three Core Web Vitals. It sits between two other metrics, helping you find where a slow load loses time:

What counts as content?#

FCP is recorded when the browser first paints any of these:

  • Text, including web font text once the font loads.
  • Images, including <img>, <image> inside SVG, and CSS background images.
  • <svg> elements.
  • Non-white <canvas> elements.

These don't count:

  • Background colours and borders: these count as First Paint, not First Contentful Paint.
  • Hidden content: elements with opacity: 0 or visibility: hidden, so pages hidden until a script runs delay FCP.
  • Invisible text: text waiting for a web font in its block period, so FCP waits for the font or the fallback.

See the Paint Timing specification for the precise rules.

First Contentful Paint vs Largest Contentful Paint#

The two metrics measure different moments in one load:

  • FCP: when the first content appears, such as a logo, navigation or heading.
  • LCP: when the largest content in the viewport appears, usually a hero image or main heading.

On a well-built page they can be the same moment. A large gap usually points to a slow hero image or content rendered late by JavaScript. If both are slow, fixing FCP often helps LCP too.

What is a good First Contentful Paint score?#

GoodNeeds improvementPoor
≤ 1.8s1.8–3.0s> 3.0s

FCP is assessed at the 75th percentile of page loads, separately for mobile and desktop. Calibre grades it against these thresholds in Synthetic, CrUX and RUM. See Core Web Vitals thresholds.

First Contentful Paint lower or equal than 1.8s are considered fast, and above 3s are considered slow.
FCP in the Lighthouse Performance score

FCP is 10% of the Lighthouse Performance score. Lighthouse scores desktop FCP more strictly than mobile, so a desktop result can look slow in Lighthouse yet be good against the thresholds above. See Lighthouse performance scoring.

How is First Contentful Paint measured?#

FCP is timed from the start of navigation, so it includes everything before the page can render:

  • Before the server responds: redirects, DNS lookup, connection setup and Time to First Byte.
  • After the HTML arrives: loading render-blocking CSS and JavaScript, then painting the first content.

In the field, Google's guidance on measuring FCP treats some page loads differently:

  • Background tabs: pages loaded in a background tab are left out.
  • Back/forward cache: restored pages count as new visits.
  • Prerendering: prerendered pages are timed from when the visitor opens them.

Use the Navigation Type filter in RUM to separate these.

What slows down First Contentful Paint?#

  • Slow server responses: nothing renders until the HTML starts arriving. Each redirect adds at least one round trip.
  • Render-blocking resources: stylesheets and synchronous scripts in the <head> must download and run before anything paints.
  • Web fonts: text can stay invisible while a font downloads.
  • Client-side rendering: an empty HTML shell shows content only after JavaScript downloads, runs and fetches data.
  • Hidden content: A/B testing anti-flicker snippets hide the page until a script loads.

How to improve First Contentful Paint#

1. Reduce server response time#

  • Improve Time to First Byte: use a CDN, caching and faster server-side rendering.
  • Avoid redirects: link straight to the final URL.
  • Use resource hints: preconnect to required origins so connections are ready when needed.

2. Eliminate render-blocking resources#

  • Inline critical CSS: inline the styles needed for the first paint, and load the rest asynchronously.
  • Defer non-critical JavaScript: use async or defer, and remove unused render-blocking scripts and CSS.
  • Render on the server: send meaningful HTML, not an empty shell waiting for JavaScript.

3. Show text while fonts load#

  • Swap fonts in: font-display: swap shows text in a fallback font while the web font loads.
  • Ship less font data: preload critical fonts, use WOFF2, and subset to the characters you need.

4. Don't hide the page#

  • Review anti-flicker snippets: limit A/B testing page-hiding to the elements under test, or shorten its timeout.

5. Monitor First Contentful Paint over time#

  • Set a budget: add an FCP performance budget in Site → Synthetic → Budgets, and subscribe to its notifications.

Track First Contentful Paint in Calibre#

  • Real User Monitoring: on the RUM Dashboard, click First Contentful Paint for its Metric Detail report. Segment by page, device, country, browser and navigation type.
  • CrUX: the 28-day, 75th percentile value from opted-in Chrome users, in Site → CrUX → First Contentful Paint. Switch to Experience Distribution for the share of good, needs improvement and poor visits.
  • Synthetic Snapshots: each Snapshot lists FCP in its key metrics and marks it on the render timeline, next to Time to First Byte and Largest Contentful Paint. Filter its Lighthouse audits to those affecting First Contentful Paint.
  • Synthetic history: track FCP over time for a page in Pulse.
A Calibre Snapshot Overview, showing the Lighthouse version used for the test, the Performance score and the Web Vitals assessment.

The Snapshot Overview shows First Contentful Paint alongside the Web Vitals for each test.

Synthetic and field FCP rarely match. A synthetic test is one load on a set device and connection. RUM and CrUX reflect your real audience's devices, networks and locations. See Why your Synthetic, CrUX and RUM numbers differ.

Browser compatibility#

Widely availableChrome 60+Edge 79+Firefox 84+Safari 14.1+

See the full browser compatibility table on MDN.

Further resources#