Skip to main content

[Deprecated] Using Statsig for experiment analysis only

While Statsig as an end-to-end (E2E) platform can be quite powerful, we understand that sometimes, you already have an assignment tool you’re feeling invested in and want to be able to simply plug and play a point solution to analyze your experiment results - Statsig’s powerful stats engine and visualization is available for you in that case, too!

Pulse is Statsig's powerful visualization that shows an experiment's impact across a wide range of metrics. It allows you to quickly assess which metrics you need to pay attention to, while finding trends across metrics, allowing you to validate existing hypotheses or devise alternative explanations. You can check out this article to learn more.

You can also play with our demo project to see Pulse in action! In this demo project, we've pre-populated a few experiments for you to interact with, so you can get a better feel how your experiment results would look on Statsig.

How to load your experiment on Statsig​

This is how data will flow from one platform to another in this scenario:

Flow

First, create a free account on Statsig, then follow these 3 easy steps to use Statsig to analyze your experiment results -

1. Connect to your data source and import exposure events (i.e. assignment data) to Statsig​

As a new user, you will see in Experiments tab that there are 2 options for you to set up your first experiment. Click on Add Data Source to import or pipe in existing experiment data.

image

Statsig provides multiple options for you to import or pipeline your exposure events -

image

Select the data source you wish to connect to, and follow the guides linked below to set up the connection and data mapping -

  • Snowflake, Databricks, Redshift, Synapse, BigQuery Data Warehouse Ingestion (most common): Follow this guide
  • Segment: Follow Segment’s instructions to set up Statsig as a destination from Segment for the source of your exposure events
  • mParticle: Follow mParticle’s instructions to set up the Statsig integration
  • API: Follow this guide

If you use LaunchDarkly for assignment, here are some resources to help you prepare your data for ingestion -

2. Generate or import the metrics you’re tracking against​

You'll need metrics to measure the impact of the the experiment you're running. There are two ways you can get your metrics into Statsig:

  • Generate events using Statsig SDK
  • Import custom events (which can be used to create custom metrics) or precomputed metrics from your data source
    • from your data warehouse: follow this guide
    • from other integrations: follow this guide

You'll be able to confirm that your imports have landed successfully (for both exposure events and metrics) through an email notification or by checking this status page under Metrics > Ingestion tab.

image

3. Set up your experiment, using exposure events and metrics that was imported in step 1 & 2​

Now when you go to create a new experiment under the Experiments tab, Statsig will automatically recognize that your experiments will be utilizing imported exposure events. During experiment setup, you will now be able to select a Foreign Experiment ID and Group ID(s) that have been extracted from your imported exposure events data. Flow group

Once your import data have landed, you'll see checks and log stream populate under Diagnostics so you can make sure your experiment was setup correctly. image

Now sit back, and let Statsig do our magic!​

And now, let Statsig do our magic! Just wait for Pulse Results to be computed (computed daily), and you can follow this guide to start analyzing your results!