Skip to main content

Real User Monitoring

The Real User Monitoring (RUM) page captures Core Web Vitals from end-user browsers via a single <script> tag. Once the beacon is installed the page shows event and session counts, p50 / p95 page-load percentiles, and a top-pages leaderboard for the last hour.


Install Snippet​

The page renders the install snippet at the top with the tenant's site key already filled in. Drop the snippet into the <head> of any HTML page to start collecting Core Web Vitals:

<script async src="…/api/v1/rum/beacon.js?site_key=default-org"></script>

The beacon is loaded asynchronously so it never blocks page rendering.


KPI Tiles​

The four headline metrics summarise the last hour of beacon events:

  • Events — Total beacon events received
  • Sessions — Distinct user sessions
  • p50 load — Median page-load duration in milliseconds
  • p95 load — 95th-percentile page-load duration in milliseconds

Top Pages​

A searchable, sortable table ranks pages by view count for the last hour:

ColumnDescription
URLPage URL — wrapped and ellipsised for long URLs
ViewsView count in the last hour
Avg loadAverage page-load time in milliseconds
p95 load95th-percentile page-load time in milliseconds

Use the global search box to slice by URL substring (e.g. /checkout); click any column to sort.


Best Practices​

  • Install the beacon on every public page — the per-page table is only useful with full coverage.
  • A p95 above two seconds usually indicates an asset or third-party-script issue rather than a server-side problem; correlate with the trace data from Tracing to confirm.
  • Compare the Sessions count here against backend traffic in Tracing to validate that the beacon is firing — a large drop means the snippet wasn't deployed.
  • Watch the URL column for unexpected entries (preview environments, internal admin pages) and exclude them at the snippet level if needed.

  • Tracing — Server-side distributed tracing that pairs with browser-side RUM
  • Monitors — Synthetic monitors that complement real-user observation