Which element is the largest contentful paint?
Whatever is biggest in the first screen when it's drawn: an image, a video's poster or first frame, a CSS background image, or a block of text. Invisible elements, full-screen backgrounds and blurry placeholders don't count. First Contentful Paint is the first thing drawn; LCP waits for the biggest.
A Lighthouse run of my five homepages (October 29, 2026) found a different element on each:
| Site | On a phone | On a desktop |
|---|---|---|
| corewebvitals.io | a portrait photo | same |
| ardennenkompas.be | the hero photo | same |
| tenshuren.be | the cover photo | same |
| coredash.app | a video's poster | same |
| marketingtracer.com | the text under the headline | the top screenshot |
So "optimize your hero image" fits three of the five. Before anyone compresses a photo, open "LCP breakdown" in PageSpeed Insights, for mobile and desktop, and see which element yours is.
What is a good LCP, and which number counts?
Good is 2.5 seconds or less and poor is over 4 seconds, at the 75th percentile of real visits: three out of four must be that fast, phones and desktops apart. At the top, PageSpeed Insights shows what real Chrome visitors got over 28 days: the number Google uses (Search Console's Core Web Vitals report shows the same data). Below it sits one load on a simulated slow phone, which can differ a lot.
That lab number moves, too. Three Lighthouse runs on coredash.app's homepage within three minutes gave 1.54, 1.55 and 2.41 seconds. The speed column in a site crawl is lab data as well.
MarketingTracer's Core Web Vitals page (Speed > Overview) shows that real-user LCP, updated weekly, next to up to six competitors, and whether the largest element was an image or text. It covers the whole site, with no number per page, and a site with too little Chrome traffic gets none.
Demo account, sample data.
How much does LCP count for Google rankings?
A little. Google says Core Web Vitals align with what its core ranking systems seek to reward and recommends good scores, without saying how much they weigh. Speed won't carry a weak answer past a better one.
In PageSpeed Insights, I look at the top block first, because that's what real visitors get. A red number below it is one test on a slow phone. If the top block is green, it costs you nothing in Google.
Why is my LCP slow?
LCP is four steps in a row, and the fix depends on which one takes the time.
It starts at the server, and slow hosting shows up in the time to first byte.
Then the browser finds the image and starts loading it. It finds it late when the image has loading="lazy", sits in a CSS background or gets added by JavaScript. Fix it with an <img> in the HTML that isn't lazy-loaded, plus fetchpriority="high". My three photo homepages have both, and Lighthouse found no delay.
Next comes the download, where a heavy file costs time; image optimization is the fix.
Last, the browser draws the element, and scripts, stylesheets and web fonts can hold that back. A text LCP has no image to download, so after the server it's all drawing: about half of this site's 1.27 seconds on a phone.
Send your developer to Largest Contentful Paint on corewebvitals.io for each step in depth.