What LCP measures
LCP is the Core Web Vital for loading. It answers "when did the visitor see the main thing on this page?" That is usually the hero image, a product photo or the headline block. Unlike older metrics such as onload, it tracks what people actually see, not when every script finished.
The elements that count are <img> elements, <image> elements inside SVG, video poster images or first frames, elements with a CSS background image, and block-level elements that contain text. The browser keeps updating the candidate as larger elements render, and stops at the first tap, scroll or key press.
Good, needs improvement and poor
| Rating | LCP |
|---|---|
| Good | 2.5 s or less |
| Needs improvement | Over 2.5 s, up to 4 s |
| Poor | Over 4 s |
Google judges a page at the 75th percentile of real page loads, split by mobile and desktop. If three in four loads of a page show the hero within 2.5 seconds, the page passes.
Example
A landing page gets 1,000 loads in a week. Sorted from fastest to slowest, load number 750 has an LCP of 3.1 s. The page's p75 LCP is 3.1 s: needs improvement. The usual culprit is a large hero image that the browser discovers late. Serving it smaller, in a modern format, and with high fetch priority often brings p75 under 2.5 s.
How to measure LCP
There are two kinds of data. Lab data comes from a test run, such as Lighthouse: repeatable and good for debugging, but one device on one network. Field data comes from real visitors, collected with the browser's performance APIs (for example Google's open-source web-vitals library) or from the Chrome UX Report. Rankings and real buyers use field data, so judge the page by it.
Common LCP mistakes
- Optimizing a lab score. A fast Lighthouse run on your laptop says little about phones on mobile networks.
- Lazy-loading the hero image. It delays exactly the element LCP measures.
- Looking at the average. Google uses the 75th percentile; the slow quarter decides.
- Assuming speed explains sales. Slow loads come with slower devices and networks, which also change how people buy.
How MetricRaven measures LCP
Turn on Page speed in Settings; it is off until you do, and it is on every plan. The tracker then loads a small add-on built on Google's web-vitals library. When a visitor leaves the page or switches tab, it sends one snapshot with LCP, INP, CLS, FCP and TTFB. Nothing is stored in the browser.
In Pages → view all, LCP and INP columns show the 75th percentile with a green, amber or red dot, next to each page's purchase rate; hover for CLS, FCP, TTFB and the number of loads. A page needs 10 loads in the period before numbers show.
Traffic → Pages → Details also lists slow pages that lose buyers, as do the API, CLI and MCP server. A visitor had a slow load when any load of the page had an LCP over 4 s or an INP over 500 ms, and a fast load when every load had an LCP of 2.5 s or less and an INP of 200 ms or less. Each page then shows the share of each group that paid and the buyers the gap suggests were lost. Treat a row as a lead, not proof. See Page speed vs conversion.
Questions
- What is a good LCP score?
- 2.5 seconds or less at the 75th percentile of page loads. Between 2.5 and 4 seconds needs improvement, and over 4 seconds is poor.
- What counts as the largest contentful element?
- The largest image, SVG image, video poster or first frame, background image, or text block visible in the viewport while the page loads. It is often the hero image or the main headline.
- Why is my LCP worse in field data than in Lighthouse?
- Lighthouse tests one simulated device. Field data covers every real visitor, including old phones and slow networks, and is judged at the 75th percentile.
- Does LCP affect conversion?
- Often, but check it on your own pages. Compare the purchase rate of visitors with fast and slow loads of the same page, and remember that device and network differ between the groups too.