How does sampling work in pageview-based DXA subscription models?

This article describes how DXA gathers data from sessions for sampling as Experience, Analysis, and Replay data in DXA accounts on pageview-based subscription models.

Note: Existing subscribers on pageview-based subscription models will be moved onto the streamlined session-based subscription model upon contract renewal. In session-based subscription models, all incoming session data is sampled as Experience data.

Gathering session data

Decibel gathers session data to different data buckets depending on the type of data being processed. The table below lists these data buckets along with the type of data each bucket processes:


Data bucket	Data type
Experience data	The data listed below is processed as Experience data: DXS Behavior metrics Segments Funnels Journeys and Discovery
Analysis data	Everything in the Experience data bucket and the data listed below is processed as Analysis data: Heatmaps Form analytics Form Discovery Technical reports
Replay data	Everything in the Experience data bucket and the data listed below is processed as Replay data: Session replays Static Heatmaps HTML searches

Sample rate

The sample rate is the percentage of session data you want DXA to process in each data bucket for sampling.

In settings > Account Settings > Properties, you can enter a sample rate for Experience data. However, you can't enter a sample rate for Analysis or Replay data, which were both set during your DXA implementation.

Note: If you'd like to increase the sample rate for Analysis and Replay data, contact your dedicated Customer Success Manager or Commercial Lead.

In the image below, the Sample Rate for Experience Data is set to 20. This means that 20% of all sessions will be sampled in the Experience data bucket.

Configuring the sample rate

Sessions gathered in the Analysis and Replay data buckets are sampled as a percentage of Experience Data. In the image above, the sample rates for both Analysis Data and Replay Data are set to 100. This means that 100% of Experience Data is sampled as Analysis Data and Replay Data.

The diagram below illustrates this process: Sampling distribution among three data buckets

In the diagram above, a property receives 10 million Sessions. 20% of these Sessions are sampled as Experience data. From these, 100% of the sessions are then sampled as Analysis and Replay data.