Auxia
  • Welcome!
  • Quick Start
    • Getting Started
      • Step 1: Ingesting your data
      • Step 2: Integrating Auxia into your product
      • Step 3: Defining your model objective
      • Step 4: Creating your first treatment
      • Step 5: Measuring success
        • Interactions
        • Engagement
  • Data Ingestion
    • Overview
      • User event data
      • User attributes
    • Data Connections
      • Google BigQuery
      • Amazon S3
      • Amplitude
        • Batch with Export API
        • Streaming with Google Pub/Sub
  • Deploying Auxia
    • Deploying with Auxia's API
      • Making your first call
      • Tracking interactions
      • Surfaces and types
      • Contextual attributes for real-time distribution
    • Delivery Integrations
      • Braze
  • API Reference
    • Get Treatments
    • Log Treatment Interactions
Powered by GitBook
On this page
  1. Quick Start
  2. Getting Started
  3. Step 5: Measuring success

Engagement

PreviousInteractionsNextOverview

Last updated 2 months ago

The image above highlights an example of the Engagement page for a typical customer. Unlike the Interactions and Adoptions dashboards, the Engagement dashboard doesn’t highlight the relative impact across treatments, but rather the consolidated impact of running Auxia as a whole across your key metrics, like your target actions or guardrail metrics.

This dashboard can help you answer questions like:

  • How is Auxia driving impact across each of my target actions (e.g. purchase completed, etc)?

  • Are there user behaviors in the product that are being negatively impacted (e.g. item likes, items saved)

The data here is comprised of the following columns:

  • Event: The event that is being ingested from your data sources

  • (Treated) Activity count: The number of times that users in the treated group have taken that action. A “Treated” user is any user that has received a treatment from Auxia.

  • (Control) Actual count: The number of times that users in the control group have taken that action. A “Control” user is any user that has not received a treatment from Auxia.

  • (Control) Normalized count: The actual count, normalized to account for differences in the size of the control group and the treated group. This is computed by multiplying the control count by the ratio of the treated group to the control group.

  • (Impact) Incremental actions: The number of treated actions minus the number of normalized control actions.

  • (Impact) Uplift vs. control: The percentage uplift of treated group over the control group.

How the dashboard is computed:

The Engagement dashboard computes daily aggregates of the number of events that occurred during the specified time period in the date range. This is how the dashboard dictates when to count an event:

  • Treated group user: The Engagement dashboard only counts the number of events that occurred if a user also received a treatment <14 days before the treatment was sent.

  • Control group user: The Engagement dashboard only counts the number of events that occurred if the user did not receive a treatment during the last 14 days.

The diagram below highlights an example of how the daily aggregates are computed.

Page cover image