This article helps you:
Identify the logic behind setting an MDE so you can more confidently run your experiment
Before running an experiment, you must estimate how you will measure success by setting the MDE (minimum detectable effect). Think of MDE as the minimum change you're hoping to see by running your experiment. Without a fail-safe calculation available for the MDE, it can be tricky to set one. With Amplitude Experiment, the default MDE is 2%; however, as the MDE is strictly linked to your unique business needs, be thoughtful during each experiment's design phase. Considerations for setting the MDE should include:
From the Plan tab in Amplitude experiment, you can choose between two experiment types: Hypothesis Testing or Do No Harm.
Let's review a case study to dive deeper into how the experiment type could alter the MDE.
The marketing director of a small arts organization is using Amplitude Experiment to help plan updates to a ticketing management system. With no data science team, the director must choose whether experiments are needed and, if so, how best to run them. The anticipated updates are:
Update #1 could benefit from a hypothesis test to ensure the placement of the "quick checkout" button boosts conversion rates by the desired amount. The marketing director notes that their next fiscal quarter goal is to increase ticket sale revenue by 3%. These company goals are important when planning for the hypothesis test, and steer the test's direction to "increase" and the MDE to 3%.
The second update, however, is an enhancement required to meet financial requirements. As a non-negotiable enhancement to the check out process, a do-no-harm experiment may help confirm that the additional payment method will not decrease completed sales for users in the checkout process. Over the last four fiscal quarters, an average of 1% of users abandoned the checkout process once starting it. Therefore, this do-no-harm experiment would have a direction set to "decrease" and an MDE set to 1%.
If running a T-test, Amplitude's duration estimator can also help set the MDE: Give yourself one month to run the experiment to get the estimated MDE.
In Amplitude, the MDE is relative to the control mean of the primary metric. For example, if the conversion rate for the control group is 10%, an MDE of 2% (0.2) would mean that a change would be detected if the rate moved outside of the range 9.8% to 10.2%.
Let's look again at the case study from the previous section. Consider how the primary metric of ticket sale purchases may require a change in the MDE:
Any unique business needs and circumstances must be considered when planning for an experiment and setting the MDE. Experiments should be run with the goal of causing as little harm as possible.
You can also set the MDE when analyzing your experiment results.
Experiments do not produce risk-free results, and running them can take a lot of time and require large sample sets, which mean higher costs and greater potential for adverse effects on users. The most important thing to remember when assessing risk is that the MDE is inversely related to sample size, meaning the smaller or more "sensitive" the MDE, the larger the sample size needed to reach statistical significance.
Here are some additional questions to ask yourself to further assess risk:
Thanks for your feedback!
May 7th, 2024
Need help? Contact Support
Visit Amplitude.com
Have a look at the Amplitude Blog
Learn more at Amplitude Academy
© 2024 Amplitude, Inc. All rights reserved. Amplitude is a registered trademark of Amplitude, Inc.