Skip to main content

User Timing

User Timing​

The Embrace SDK automatically captures custom performance marks and measures created by your application via performance.mark() and performance.measure(), giving you visibility into your own timing instrumentation without any extra backend work.

How User Timing Instrumentation Works​

The SDK observes PerformanceMark and PerformanceMeasure entries via a PerformanceObserver. Marks are emitted as logs and measures are emitted as spans, reflecting their different shapes: a mark is a discrete point in time, while a measure has a defined start and duration.

To keep data volumes manageable, the SDK deduplicates entries by name per page URL — only the first occurrence of a given name is captured per navigation. A cap of 200 marks and 100 measures per page is applied after deduplication.

Data Captured​

For each mark, the SDK captures:

  • The mark name and timestamp
  • The entry type (mark)
  • Any detail object provided via performance.mark(), serialized as the log body

For each measure, the SDK captures:

  • The measure name (used as the span name) and start time
  • The duration
  • The entry type (measure)
  • Any detail object provided via performance.measure(), attached as a span attribute

Filtering Entries​

You can limit which entries are captured using the allowedEntries option, which accepts either an array of names or a predicate function:

import { initSDK } from '@embrace-io/web-sdk';

initSDK({
  appID: "YOUR_EMBRACE_APP_ID",
  appVersion: "YOUR_APP_VERSION",
  defaultInstrumentationConfig: {
    'user-timing': {
      allowedEntries: ['my-app-measure', 'checkout-flow'],
    },
  },
});

Integration with Other Features​

  • User timing marks and measures are associated with the current session and page
  • Custom dashboards can be built to monitor percentile durations and start times, filtered by name, entry type, or detail attributes
  • Alerts can be set on measure durations exceeding specific thresholds