Planning and Launching Experiments

Cette page n'est pas encore disponible en français, sa traduction est en cours.
Si vous avez des questions ou des retours sur notre projet de traduction actuel, n'hésitez pas à nous contacter.

Overview

Datadog Experimentation allows you to measure the causal relationship that new experiences and features have on user outcomes. To do this, Experimentation randomly allocates traffic between two or more variations, using one of the variations as a control group.

This page walks you through planning and launching your experiments.

Setup

To create, configure, and launch your experiment, complete the following steps:

Step 1 - Create your experiment

  1. Navigate to the Experiments page in Datadog Product Analytics.
  2. Click + Create Experiment.
  3. Enter your experiment name and hypothesis.
create an experiment and add a hypothesis for the experiment.

Step 2 - Add metrics

After you’ve created an experiment, add your primary metric and optional guardrails. See Defining Metrics for details on how to create metrics.

create an experiment and add a hypothesis for the experiment.

Add a sample size calculation (optional)

After selecting your experiment’s metrics, use the optional sample size calculator to determine how small of a change your experiment can reliably detect with your current sample size.

  1. Select the entry point of your experiment. This specifies when in the user journey they will be enrolled into the test. For example, if you plan to run an experiment on users who visit the homepage, select the homepage view as your entry point.

  2. Click Run calculation to see the Minimum Detectable Effects (MDE) your experiment has on your metrics. The MDE is the smallest difference that you are able to detect between your experiment’s variants.

Sleect an entrypoint event to run a sample size calculation

Step 3 - Launch your experiment

After specifying your metrics, you can launch your experiment.

  1. Select a Feature Flag that captures the variants you want to test. If you have not yet created a feature flag, see the Getting Started with Feature Flags page.

  2. Click Set up experiment on feature flag to specify how you want to roll out your experiment. You can either launch the experiment to all traffic, or schedule a gradual rollout.

Set up an experiment on a Feature Flag.

Next steps

  1. Defining metrics: Define the metrics you want to measure during your experimentation.
  2. Reading Experiment Results: Review and explore your Experiment results.
  3. Learn more about Minimum Detectable Effects: Choose an appropriately sized MDE.

Further reading

Documentation, liens et articles supplémentaires utiles:

Make data-driven design decisions with Product AnalyticsBLOG more more Defining Experiment MetricsDOCUMENTATION more more