When Dynatrace ingests RUM data, it adds fields to user events and user sessions that your frontends don't send themselves. These built-in enrichments make your data ready for analysis without any configuration—for example, by rating how good your users' experience was. This page explains which enrichments Dynatrace applies and how they're calculated.
To add your own business or application context instead, use event and session properties.
Experience Score is a single, weighted score between 0 and 1 that summarizes how good your users' experience was, both for a frontend as a whole and within individual sessions.
Dynatrace classifies every relevant user event as satisfied, tolerating, or frustrated, and then combines these classifications into one score per frontend and one score per session. Because the classification runs in OpenPipeline, the team responsible for a frontend can decide what counts as a good or bad experience. By default, Dynatrace scores every user action by its duration. You can extend this to give business-critical user actions a higher weight, judge web user actions by Core Web Vitals, or treat errors and crashes as frustrating. For details, see Customize Experience Score with OpenPipeline.
Experience Score replaces Apdex ratings and the user experience score from RUM Classic.
The built-in RUM Agent ingest source in OpenPipeline includes the Experience Score calculation pre-processing. The pre-processing runs on all user action events and adds two fields to each one:
experience.level: The classification of the event, either satisfied, tolerating, or frustrated.experience.weight: An integer that defines how much the event counts toward the score, relative to other events.For the full specification of these fields, see User actions in the Semantic Dictionary.
By default, the pre-processing classifies user actions by their duration and gives every user action the same weight:
| User action duration | experience.level | experience.weight |
|---|---|---|
Less than 2 seconds |
| 1 |
2 seconds to less than 5 seconds |
| 1 |
5 seconds or more |
| 1 |
To see the processors that make up the pre-processing, open the RUM Agent ingest source in
Settings > Process and contextualize > OpenPipeline > User events > Ingest sources.
Only events that carry experience.level count toward the score. Events that no processor classifies are excluded. Scoring only classified events lets you opt specific event types in or out through OpenPipeline instead of relying on a fixed scope.
The Experience Score is the weighted average of all classified events:
Experience Score = sum(level_value × experience.weight) / sum(experience.weight)
The level values are:
satisfied = 1tolerating = 0.5frustrated = 0With the default classification and equal weights, this formula matches the Apdex calculation.
The resulting score maps to the following bands:
| Band | Experience Score |
|---|---|
Excellent | 0.94 or higher |
Good | 0.85 to less than 0.94 |
Fair | 0.70 to less than 0.85 |
Poor | 0.50 to less than 0.70 |
Unacceptable | Less than 0.50 |
For example, a session with two satisfied user actions and one frustrated user action, all with a weight of 1, has an Experience Score of (1 + 1 + 0) / 3 = 0.67, which is in the Poor band.
You can get the Experience Score for frontends and for sessions:
dt.frontend.experience.weight metric. For details, see Built-in RUM metrics.experience.score on the user.sessions record, together with the matching band in experience.score.band. You can query both with DQL, as described in Customize Experience Score with OpenPipeline, and filter sessions by band in
Users & Sessions.
Users & Sessions and review what happened in them.