Introduction

​

Use Case Walkthrus

​

Website Checkout A/B Test with Apple Pay

Running experiments against e-commerce checkout flows is a common practice, used to improve the efficiency and effectiveness of the checkout process for users. Today, we walk through an example using Winning Variant on Snowflake AI Data Cloud.

Watch on YouTube

​

Split Testing ML Model Versions

When developing a new ML model, before putting it out into the wild, it’s important to test its performance against real business metrics. Not just historical testing (what might have happened if we had this version before), but in real time against live activity.

Winning Variant allows you to run a split test comparing multiple model versions at the same time against live traffic. The results will provide an objective measure of change of the business metric we care about, and give a more reliable signal we can use to determine if the new model is “better” than the old.

Watch on YouTube

​

Snowflake Native App Setup

Install and configure the Winning Variant Native App in your Snowflake account.

​

Manage Experimentation

Create labs, subject types, and experiments in your new experimentation engine.

​

Experimention Data

Learn about the raw experimentation data you can not activate for better, deeper analysis.