Công cụ kiểm tra PageSpeed & Lighthouse miễn phí

It runs one Lighthouse lab audit in a simulated environment and reports category scores, lab timing metrics and prioritized diagnostics. Field INP appears only when enough real-user data is available.

Lighthouse scores run from 0 to 100. Broadly, 90 and above is good, 50 to 89 needs improvement, and below 50 is poor. Treat it as lab guidance, not a real-user guarantee.

Lighthouse runs one audit under simulated network and device conditions, so it may not match real visitors' varied hardware and connections. Lab data is reproducible and diagnostic; field data reflects actual user experience over time.

Work through the listed opportunities, commonly compressing and sizing images, deferring unused scripts, eliminating render-blocking resources, and enabling caching. Each opportunity estimates potential savings, so prioritize the ones with the largest estimated time reduction.

PageSpeed runs a Lighthouse lab audit with simulated conditions and diagnostic opportunities. The Core Web Vitals Checker reports real-user field data at the 75th percentile. Use lab to diagnose, field to judge actual experience.

LCP, CLS, FCP and TTFB on this panel come from the single simulated audit. INP is not produced by that lab run; it appears only when the accompanying field dataset has enough real-user observations.

Lighthouse simulates a run each time, so variability in network modeling, server response and page scripts shifts scores. Run several audits and look at the trend rather than treating one number as absolute.

No. Lighthouse checks automatable accessibility issues only and cannot verify everything, like meaningful alt text or keyboard flows. A high score is a helpful baseline, not proof of full accessibility compliance.

A lab audit exposes reproducible loading, accessibility and implementation diagnostics. Use it to find technical work, then verify the outcome with field data; one lab run does not predict rankings or every visitor's experience.

No. It is one lab audit in a simulated environment representing a single test condition. To understand real visitors across devices and networks, pair it with real-user field data from the Core Web Vitals Checker.