Skip to content
2.2Intermediate8 min

LCP Optimization: The Hardest Core Web Vital

Lucas Blochberger
Definition

Largest Contentful Paint (LCP) measures the time until the largest visible content block is fully rendered. With only 62 percent mobile pass rate, LCP is the hardest Core Web Vital and the primary bottleneck.

Key Takeaways

  • LCP target is under 2.5 seconds, measured at the 75th percentile of real user data (CrUX); 2.5 to 4.0 seconds is considered needs improvement, over 4.0 seconds is poor.
  • LCP is the hardest Core Web Vitals metric to pass: only 62 percent of mobile pages achieve a good score, compared to 77 percent for INP and 81 percent for CLS.
  • On around 85 percent of desktop and 76 percent of mobile pages, an image is the LCP element; the most common LCP elements are hero images, large text blocks, and videos.
  • LCP breaks down into four subparts: Time to First Byte, Resource Load Delay, Resource Load Time, and Element Render Delay. Diagnostics show where time is actually lost.
  • The LCP image must never be lazy-loaded; modern formats (WebP/AVIF), fetchpriority=high and preload, plus a fast server (TTFB) are the most effective levers.
  • Field and lab data complement each other: Lighthouse shows the cause, CrUX/PageSpeed/Search Console show whether real users are affected. An optimization only counts once it shows up in field data.
  • Fast LCP measurably contributes to conversion and revenue and acts as a ranking tiebreaker; for Austria with around 44 percent mobile traffic, mobile LCP is the most important focus.

LCP is often the easiest Core Web Vital to improve, as optimizations are clearly defined.

The Four LCP Sub-Categories

Google divides LCP into four phases: Time to First Byte (TTFB, server response time), Resource Load Delay (delay until the LCP resource download starts), Resource Load Duration (download time of the LCP resource), and Element Render Delay (time to render after download).

Optimization Strategies

Server optimization: Bring TTFB below 800ms. Upgrade hosting, enable caching, use a CDN. A CDN can improve LCP by 30-50 percent.

Image optimization: AVIF as primary format (50% smaller than JPEG), responsive images with srcset, preload for the LCP image via link rel=preload, lazy loading only for images below the fold.

Render optimization: Inline critical CSS, eliminate or defer render-blocking JavaScript, preload fonts with link rel=preload.

LCP and AI Crawlers

AI crawlers typically wait only 1-5 seconds for a page. An LCP above 2.5 seconds means AI systems may see an incompletely loaded version of the page.

Data & Statistics

62 percent of mobile and 74 percent of desktop pages achieve a good LCP value (under 2.5 s); mobile shows almost twice as many poor LCP values with 13 percent compared to 7 percent. LCP has the lowest mobile pass rate of the three CWV at 62 percent (INP 77%, CLS 81%).

HTTP Archive Web Almanac 2025, Performance chapter (CrUX-Daten Juli 2025) [international] (2025)

Only 48 percent of mobile and 56 percent of desktop origins pass all three Core Web Vitals; LCP is the limiting metric.

HTTP Archive Web Almanac 2025, Performance chapter [international] (2025)

For 85.3 percent of desktop and 76 percent of mobile pages, an image is the LCP element; text-based LCP elements account for 23.7 percent on mobile.

HTTP Archive Web Almanac 2025, Performance chapter (LCP content types) [international] (2025)

Around 16 to 17 percent of pages lazy-load their LCP image (stable since 2024); native loading=lazy usage increased on mobile from 9.5 to 10.4 percent.

HTTP Archive Web Almanac 2025, Performance chapter (LCP lazy loading) [international] (2025)

JPG is the most common LCP image format at around 57 percent (minus 4 percentage points compared to 2024), WebP follows with 11 percent (plus 4 PP), AVIF reaches 0.7 percent, PNG 26 percent.

HTTP Archive Web Almanac 2025, Performance chapter (LCP image formats) [international] (2025)

A 31 percent better LCP led to 8 percent more sales, a 15 percent better lead-to-visit rate, and an 11 percent better cart-to-visit rate (A/B test, optimized page 5.7 s vs. 8.3 s).

web.dev (Google) - Vodafone case study [international] (2021)

In the retail segment, a 0.1 second faster mobile load time increased the conversion rate by 8.4 percent, the average order value by 9.2 percent, and page views per session by 5.2 percent.

Deloitte Digital / Fifty-Five for Google - 'Milliseconds Make Millions' report (p. 17, Retail) [international] (2020)

In September 2025, around 44.4 percent of all internet page views in Austria [Austria] were on mobile devices such as smartphones (excluding tablets, around 1.9 percent).

Statista (data from StatCounter) - Share of mobile devices in all page views in Austria [Austria] (2025)
Summarize with AIChatGPTClaudePerplexity

Opens the chat with a prepared prompt.

FAQ

What is a good LCP value?
A good LCP value is under 2.5 seconds. Google defines three thresholds: good (under 2.5 s), needs improvement (2.5 to 4.0 s), and poor (over 4.0 s). What matters is the 75th percentile of real user data (CrUX), meaning 75 percent of actual page loads must meet the threshold. A good lab value alone is not enough if slower devices and connections push the percentile upward.
How can I optimize LCP?
Address the four subparts. Lower TTFB through server-side rendering, caching, CDN, HTTP/3, and Brotli. Deliver the LCP image in modern formats (WebP/AVIF) with correct dimensions, srcset/sizes, fetchpriority=high and preload, and never lazy-load it. Eliminate render-blocking CSS and JavaScript through critical inline CSS, defer/async, and font-display: swap with preload for web fonts.
What is the LCP element on a page?
The LCP element is the largest visible content block in the viewport. Typically these are hero images, large text blocks, or videos. In practice, images dominate: according to the HTTP Archive Web Almanac 2025, an image is the LCP element on around 85 percent of desktop and 76 percent of mobile pages. Which element specifically counts is shown by tools like PageSpeed Insights or Chrome DevTools.
Why is my LCP poor despite a fast image?
Because LCP consists of four subparts and the image affects only one of them. Often the bottleneck is at the server (high TTFB), in late discovery of the image (Resource Load Delay), or in delayed rendering due to JavaScript and CSS (Element Render Delay). In modern frameworks, the LCP element may be fully loaded but delayed in rendering by hydration. A subpart analysis shows where time is actually lost.
What is the difference between field and lab data for LCP?
Field data comes from real users and determines the official 75th percentile. Sources are CrUX, PageSpeed Insights, and Google Search Console. Lab data is generated synthetically under controlled conditions, such as with Lighthouse, and is suitable for testing before deployment and isolating causes. Both are necessary: lab data shows the cause, field data shows whether real users are affected. An optimization only counts once it shows up in field data.
Should the LCP image be lazy-loaded?
No. The LCP image must never be lazy-loaded, because this delays the start of loading and worsens the LCP score. According to the HTTP Archive Web Almanac 2025, around 16 to 17 percent of pages incorrectly lazy-load their LCP image. loading=lazy belongs exclusively on images below the fold. The LCP image should instead receive fetchpriority=high and a preload; in Next.js, the priority prop of next/image does exactly that.
How does LCP help for AI crawlers and GEO?
Fast server-side rendering, low TTFB, and minimal render-blocking JavaScript make content reliably accessible to AI crawlers and search engine bots as well. The same architecture that improves LCP facilitates delivery to AI crawlers. Important is the distinction: LCP is a performance metric for user experience, not a direct GEO factor. Generative Engine Optimization is determined by content, structure, and citability, not by load time. LCP improves accessibility but does not replace GEO measures.

How does your website perform?

Get a free, AI-powered SEO report of your website by email: technical SEO, on-page, keywords & competitors. No obligation.

Get a free SEO audit →

Or go straight to SEO Agency in Vienna

Previous← Core Web Vitals: LCP, INP and CLS ExplainedNextINP Optimization: Measuring and Improving Interactivity →