How do I check Core Web Vitals for real visitors?

Short answer

Open Monitoring › Performance. It shows Largest Contentful Paint, Interaction to Next Paint, Cumulative Layout Shift, First Contentful Paint and Time to First Byte, measured in real visitors' browsers. Read P75 first, then open Slowest pages and break down by page, device or country to find what to fix.

Collection is on by default in the website script; nothing extra to install.

Thresholds

Values in between need improvement. The first three are the Core Web Vitals; the 0–100 score on the page combines only those three.

Read the report

1. Cards: switch between P95, P75, median and average. P75 is closer to slower visitors and matches how Core Web Vitals are usually judged.
2. Slowest pages: sorted by P75 LCP.
3. Breakdown: by page, country, device, browser or operating system, at P50, P75, P90 or P99.
4. Load timing: splits a load into backend (TTFB), network transfer and frontend render, so you can tell whether the server or the page is slow.

Then check Web analytics › Pages to see whether the slow pages are also losing visitors.

Why is it empty or patchy?

• The script tag sets data-vitals="false", or data-auto-track="false", which turns off automatic collection including vitals.
• An old snippet that does not collect vitals; replace it with the current one.
• The script is blocked for some visitors; TapCub does not substitute server timings.
• Few samples on a page make percentiles jump; widen the date range.

Vitals are sent once per page view, when the first page is hidden.

Common questions

Why do these numbers differ from a lab test? Lab tools run one load on one device. This report collects real loads from all visitors, on their devices and networks.

Is there a previous-period comparison? No; compare by changing the date range.

Where are JavaScript errors? On a separate page; see find JavaScript errors.

Background: Core Web Vitals, Largest Contentful Paint and good page load time.