Largest Contentful Paint (LCP)

What LCP measures

Largest Contentful Paint records how long it takes for the largest visible element in the viewport to finish rendering. On most business sites that is the hero image, the main headline, or a background video poster frame.

It is a deliberately blunt instrument, and that is the point. Earlier metrics measured when a page started doing something — LCP measures when the visitor can actually see the thing they came for.

Thresholds

• Good: 2.5 seconds or less
• Needs improvement: 2.5 to 4.0 seconds
• Poor: over 4.0 seconds

Measured at the 75th percentile of real visits, split between mobile and desktop.

What the timer is actually waiting on

LCP is rarely one slow thing. It is usually a chain, and the chain has to complete in order:

1. Server response. How long before the first byte arrives. A slow host or an overloaded database puts everything else behind schedule before the browser has done anything.
2. Render-blocking resources. Stylesheets and scripts the browser must finish processing before it will paint anything at all.
3. Resource load time. Downloading the hero image itself — the single biggest lever on most small business sites.
4. Render delay. The gap between the image arriving and it appearing, often caused by fonts, JavaScript, or lazy-loading applied to the wrong element.

The mistakes we see most

Enormous hero images. A photo straight off a modern phone can be 4–8 MB. Resized to its actual display size and served as WebP or AVIF, the same image is often under 200 KB with no visible difference.

Lazy-loading the hero. Lazy loading is genuinely useful — for images below the fold. Applied to the above-the-fold hero, it forces the browser to wait before even requesting the one image LCP is timing. This is a surprisingly common plugin default.

Fonts blocking text. If your headline is the LCP element and the font loads from a third-party server, LCP waits on that request. Self-hosting fonts and using font-display: swap removes the dependency.

No CDN. A visitor in Morehead City hitting a server in Oregon pays for that distance on every request.

Why it is usually the cheapest metric to fix

Of the three Core Web Vitals, LCP responds best to straightforward work. Compressing and correctly sizing the hero image, removing a render-blocking script, and putting the site behind a CDN routinely takes a page from four seconds to under two — with no redesign and no rebuild.

INP and CLS problems are more often structural. LCP problems are usually just an oversized file nobody looked at.