Performance
The numbers come from real loads in the visitor’s browser, not from a lab run. Collection switches are in Enhanced measurement.
ConsoleMonitoringPerformanceWhat the report shows
Five cards match the five metrics. The thresholds are:
| Metric | Good (at most) | Poor (above) |
|---|---|---|
| Largest contentful paint | 2.5 seconds | 4 seconds |
| Cumulative layout shift | 0.1 | 0.25 |
| Interaction to next paint | 200 milliseconds | 500 milliseconds |
| First contentful paint | 1.8 seconds | 3 seconds |
| Time to first byte | 800 milliseconds | 1.8 seconds |
Anything in between is needs improvement.
Cards can switch among P95, P75, median, and average. The trend and the breakdown can use P50, P75, P90, or P99. Breakdown dimensions are page, country, device, browser, and operating system. Load timing splits one load into backend (time to first byte), network transfer, and frontend render (response end until the document is complete). A Slowest pages list is sorted by P75 largest contentful paint, and a 0–100 score combines largest contentful paint, interaction to next paint, and cumulative layout shift.
How to use it
Start with P75 and the share rated good, then break down by page and find the slow paths. Check Pages to see whether those pages are also losing people.
Limits
Metrics are reported by the visitor. If the script is blocked, or an old snippet does not collect them, the page is empty; server time is not substituted. With no samples, the page says you need the current analytics script. Percentiles jump when a page has few samples. The score only combines the three core metrics. It is not a full-site audit. This page has no previous-period comparison.
FAQ
Is this the same as the errors page? No. This page is load time. Errors & frustration is script exceptions and frustrated clicks. A slow page does not have to have an error.
Median or P75? The median is the middle of the samples. P75 is closer to the slower visitors. The slowest-pages list is sorted by P75.
Errors & frustration
The install snippet has to turn error collection on, which loads a small extra script. If it is off, a notice at the top points you at the tracking code so you can enable it and update the snippet.
ConsoleMonitoringErrorsWhat the report shows
The cards are error count, error rate, affected visitors, and affected sessions. The rate is error count divided by pageviews in the selected range. The trend is error count over time. Each group shows count, sessions involved, and last seen. Status is Active, Resolved, or Ignored. Expand a group for each occurrence: time, page, place, device, and session.
Under the groups is the frustration table:
- Rage click: at least 3 clicks on the same element within 500 milliseconds.
- Error click: a script error within 1 second after the click.
- Dead click: the page did not change after the click. The current aggregate always stores 0 in this column; collection does not write it yet.
A page is listed only if it has at least 20 pageviews and a rage click or error click greater than zero. The per-thousand rate is rage clicks plus error clicks, divided by pageviews, times 1000.
How to use it
Mark a finished group Resolved or Ignored to take it out of Active. Open session goes to Sessions log. Click position on the element is in Heatmaps & experience.
Limits
If error collection is off, an empty group list does not mean production threw nothing. Frustration clicks depend on the autocapture script. If that is off, the table says so. A zero dead-click column does not mean there were no dead clicks. This page has no previous-period comparison.
FAQ
Is the error rate the same as script errors on the data report? Both counts are script-error events. The rate on this page is count divided by pageviews. The data report shows the error count itself. See Web metric definitions.
Do rage clicks become a funnel step? Not automatically. Funnels are in Basic funnels, paths & retention. Heatmaps can show rage clicks and ordinary clicks on the same page. See Heatmaps & experience.