Skip to main content

Running an A/A (aa) Test

In this guide, we will walk you through how to leverage Statsig’s platform to run an A/A test on your product.

This guide assumes that you have successfully set up and configured your Statsig SDK. For a step-by-step guide on how to do this, see our “Your first feature” guide.

Why run an A/A (aa) test?

There are many reasons to run an A/A test, one of the most common being to validate a new experimentation engine you may be integrating with (in this case Statsig). For new users just getting started with Statsig, we often recommend running an A/A test to provide a “low-stakes” first test environment, ensuring that you’ve got your metrics set up correctly and are seeing exposures flowing through as expected before kicking off your first real A/B test.

Here at Statsig, we are continuously running both live and offline simulation A/A tests on our stats engine- if you want to check one out for yourself, see this example in our Swaggr Demo Project!

How to run an A/A test

The easiest way to run an A/A test in Statsig is by leveraging a Feature Flag. You can also leverage an Experiment to run an A/A, but we chose to use a Feature Gate for this tutorial for simplicity.

Step 1: Create a new feature gate in the Statsig console

Log into the Statsig console at https://console.statsig.com/ and navigate to Feature Gates in the left-hand navigation panel.

Click on the Create button and enter the name and (optional) description for your feature gate. We will call our feature gate “aatest_example”. Click Create.

create_new_fg_empty

In the Setup tab, define the rules for this feature gate. Tap + Add New Rule. While you could run an A/A test on a specific user-group, platform, etc. the easiest setup is to simply divide all of your traffic 50/50 and deliver the same experience (your default product experience) to each group.

add_new_rule_empty

To do this, under Criteria select Everyone (you may need to scroll up), name your rule, and then change the Pass Percentage to 50%. Click Add Rule and that’s it! Tap Save Changes in the upper right-hand corner.

add_new_rule_filled

Your feature gate setup should now look as follows-

aa_rule_filled_out

Check that it is working as expected by typing in some dummy user IDs into the console- roughly 50% of the time your IDs should pass, and 50% of the time they should fail.

check_rule_pass

check_rule_fail

Step 2: Check the feature gate in your application and log an event

Copy the code snippet in the upper right hand corner of your feature gate page under the < > symbol and drop it into your application at the point you want to call the A/A check.

statsig.checkGate("aatest_example") 

Now when a user renders this page in their client application, you will automatically start to see a live log stream in the Statsig console when you navigate to the Diagnostics tab for your feature gate.

logstream

Step 3: Review A/A test results

Within 24 hours of starting your experiment, you'll see the cumulative exposures in the Pulse Results tab of your feature gate.

cumulative_exposures

This will break down your logged exposures (as well as the distribution of the logged exposures). If something looks off, check the Diagnostics tab for more granular, day-by-day exposure breakdowns at both the Checks and User level.

In the Metric Lifts panel, you can see the full picture of how all your tagged metrics are performing.

pulse_results_empty

What should you expect to see?

  • Exposures- make sure you’re seeing exposures flowing through as expected from your product. If you’re not seeing exposures, use the Diagnostics tab and the Exposure Stream to debug
  • Pulse results- roughly 5% of your metrics in Pulse should be showing a statistically significant change due to the 95% confidence interval of Statsig’s stats engine

We recommend running your A/A long enough to reach most of your weekly active users, or at least a week.