
Key points
- LCP is good at 2.5 seconds or less, and poor above 4 seconds, measured at the 75th percentile of visits.
- Find the LCP element first. Fixes differ completely for an image, a text block and a background image.
- Break LCP into four parts (server, discovery, download, render) and fix the largest one.
What LCP measures
Largest Contentful Paint reports the time from the start of navigation until the largest image or block of text inside the viewport has finished rendering. The browser keeps updating its candidate while the page loads and stops when the visitor interacts with the page.
The element types that can be an LCP candidate are images (<img>, and images inside SVG), video poster images, elements with a background image loaded through url(), and block-level elements containing text. Anything outside the initial viewport is ignored.
| Rating | LCP |
|---|---|
| Good | ≤ 2.5 s |
| Needs improvement | 2.5–4 s |
| Poor | > 4 s |
Find your LCP element
You cannot fix LCP without knowing what it is. In a WebsiteHealthLabs report the Performance section names the LCP element when Lighthouse reports one. Chrome DevTools shows the same under Performance → Timings → LCP, and Lighthouse reports it in the “Largest Contentful Paint element” diagnostic.
Once you know the element, the likely causes narrow quickly:
- A hero image: too large, wrong format, found late, or lazy-loaded.
- A text block: usually blocked by a slow web font or by render-blocking CSS.
- A CSS background image: invisible to the browser's preload scanner until the CSS has been downloaded and parsed.
The four parts of LCP
Every LCP can be split into four sequential parts. Google's web.dev guidance suggests most of the time should go to the first and third, and very little to the second and fourth.
- Time to first byte (TTFB): from navigation until the HTML starts to arrive.
- Resource load delay: the gap between the HTML arriving and the browser starting to download the LCP resource. Should be close to zero.
- Resource load duration: the download of the LCP resource itself.
- Element render delay: the gap between the resource finishing and the element actually painting. Blocking scripts or CSS cause it.
If your delay is in part 2 or 4, shrinking the image will not help. Fix the discovery or the blocking instead.
How to improve LCP
Make the image cheap and early
- Serve a modern format (AVIF or WebP) at the size it is displayed, with
srcsetandsizesfor different screens. - Put the image in the HTML as an
<img>rather than a CSS background, so the preload scanner finds it immediately. - Add
fetchpriority="high"to the LCP image and never putloading="lazy"on it.
<img src="/img/hero-1200.avif"
srcset="/img/hero-600.avif 600w, /img/hero-1200.avif 1200w"
sizes="(min-width: 1200px) 1200px, 100vw"
width="1200" height="630"
fetchpriority="high"
alt="Product dashboard overview">
Speed up the server response
- Cache rendered HTML where you can (page cache, CDN).
- Avoid redirect chains before the final page; each hop adds a round trip.
- Reduce slow database queries and blocking work during page generation.
Remove render-blocking work
- Inline the small amount of CSS needed for the first screen and load the rest without blocking.
- Add
deferto scripts that do not need to run before first paint. - For text LCP, use
font-display: swap(oroptional) and preload the one font file the headline needs.
One change at a time. Re-run the audit after each fix. LCP in a lab run varies a little between runs, so look for clear improvements rather than a few milliseconds.
See also: Page weight explained, Redirects: 301 vs 302.
FAQ
Why does LCP differ between mobile and desktop?
Mobile tests use a slower simulated network and CPU, and the viewport is smaller so a different element can be the largest. A site can pass on desktop and fail on mobile with the same code.
Should I lazy-load images?
Lazy-load images that are below the fold. Never lazy-load the image that is your LCP element; it delays the very thing LCP measures.
Does a CDN fix LCP?
It helps when the delay is in server response or download time because content is served closer to the visitor. It does not help if the LCP image is discovered late or blocked by scripts. Find the slow part first.
Is LCP the same as page load time?
No. LCP is when the largest visible element appears. The load event can fire much later (or earlier), because it also waits for resources that are not visible.