Core Web Vitals for beginners: measure and improve LCP, INP and CLS
Start with the experience your visitors have, then use the right measurement to decide what to fix.
Tools
Core Web Vitals thresholds: LCP, INP and CLS reference
Read the metric boundaries alongside the measurement conditions that make them meaningful.
Largest Contentful Paint (LCP)
Definition
The render time of the largest image, video frame or text block visible in the viewport, measured from when the page starts loading.
How it is measured
Reported by Chrome for real users (field) and by Lighthouse (lab). The 75th percentile of page loads is what counts.
CrUX field results use the 75th percentile over a rolling 28-day collection period; individual lab runs are different.
Thresholds
- Good Good: 2.5 s or less
- Needs improvement Needs improvement: above 2.5 s and up to 4.0 s
- Poor Poor: more than 4.0 s
How to improve it
- Make the likely LCP content discoverable in the initial HTML where possible.
- Do not lazy-load the LCP image; use fetchpriority="high" selectively for the important initial image.
- Reduce server response time and remove render-blocking CSS and JavaScript.
- Use responsive, compressed images in modern formats.
Interaction to Next Paint (INP)
Definition
A near-worst interaction latency across the visit, accounting for outliers on highly interactive pages, from input until the next frame is painted.
How it is measured
Field data describes real visits. Controlled interactive browser tests can diagnose interaction latency; a standard Lighthouse navigation run reports TBT rather than field INP.
CrUX field results use the 75th percentile over a rolling 28-day collection period; individual lab runs are different.
Thresholds
- Good Good: 200 ms or less
- Needs improvement Needs improvement: above 200 ms and up to 500 ms
- Poor Poor: more than 500 ms
How to improve it
- Break long JavaScript tasks into smaller pieces and yield to the main thread.
- Reduce the work done in event handlers; defer anything the user does not need to see immediately.
- Audit third-party scripts and tag managers.
- Keep the DOM small and avoid forced synchronous layouts.
Cumulative Layout Shift (CLS)
Definition
A score for how much visible content moves unexpectedly during the page's lifetime, based on the largest burst of shifts.
How it is measured
Reported in field and lab tools. Unitless: 0 means no qualifying layout shift was recorded.
CrUX field results use the 75th percentile over a rolling 28-day collection period; individual lab runs are different.
Thresholds
- Good Good: 0.1 or less
- Needs improvement Needs improvement: above 0.1 and up to 0.25
- Poor Poor: more than 0.25
How to improve it
- Give every image, video and iframe explicit width and height or an aspect-ratio.
- Reserve space for ads, embeds and banners before they load.
- Avoid inserting content above existing content unless the user asked for it.
- Use transform for animations instead of properties that trigger layout.
Thresholds reflect Google's published guidance. Check the linked documentation for changes. web.dev
How to use this tool
- Identify the metric and its unit in your source report.
- Check whether the value is field data at the 75th percentile or a result from an individual lab test.
- Compare it with the good, needs-improvement and poor ranges shown here.
- Follow the relevant guide to investigate the cause of a value outside the good range.
Keep mobile and desktop results separate. Also check whether the report describes one URL or an origin that combines many pages. Thresholds can classify either value, but the scope determines what you can conclude about an individual page.
About the results
LCP measures loading of the largest eligible visible content. Good is 2.5 seconds or less; above 2.5 and up to 4.0 seconds needs improvement; above 4.0 seconds is poor.
INP measures responsiveness to qualifying interactions. Good is 200 milliseconds or less; above 200 and up to 500 milliseconds needs improvement; above 500 milliseconds is poor.
CLS measures unexpected layout movement and has no time unit. Good is 0.1 or less; above 0.1 and up to 0.25 needs improvement; above 0.25 is poor. Do not append seconds or milliseconds to a CLS value.
These ranges reflect Google's published Core Web Vitals guidance. They are not the same scale as a Lighthouse performance score from 0 to 100. The beginner guide explains how the metrics and tools fit together.
What does the 75th percentile mean?
The p75 value is at or above roughly three quarters of the measured values in the relevant population. It is not the mean and not your fastest test. The field assessment uses this approach to evaluate whether a broad share of visits meets the targets.
A single lab test can be useful for diagnosis, but it does not establish a population percentile. Use the field versus lab guide before comparing reports that use different collection methods or time windows.
What if one metric is missing?
Missing data should remain missing. A visit without a qualifying interaction does not provide an INP value, and a page may lack enough eligible field samples for reporting.
Read the source tool's coverage rules. PageSpeed Insights can assess good LCP and CLS when INP data is insufficient; without sufficient LCP or CLS, it cannot make that assessment. That rule does not authorize inventing an INP result or transferring another page's value to this one.
What should I do near a threshold?
Keep the actual values and distributions, not just the color. A small regression can move a result near a boundary into the next range, while rounding can make two values look more similar than they are. Investigate representative pages and repeat comparable tests.
Use the Core Web Vitals checker for values you have already measured, then follow the relevant implementation guide. A threshold helps choose priorities; it is not a promise of rankings or proof that every visitor has a good experience.
Related guides
Field data vs lab data: why PageSpeed Insights results differ
Two results can disagree without either being wrong. Check the population, time window and measurement before choosing a fix.
Do Core Web Vitals affect Google rankings? What Google documents
Core Web Vitals matter to Google Search, but a passing result is not a ranking guarantee or a substitute for useful content.