If you didn’t get a chance to place your swag order the store is now OPEN! Please use the link here to get to the online swag store. Quantities are very limited so be sure to get your swag before it runs out!
Departments across Overstock conduct experiments to test new product features and algorithm changes before they go into production. These are typically A/B tests performed on live traffic and rely on statistical analysis to determine whether the observed difference is real or merely reflect random fluctuations in the data. There are many practical challenges, such as measuring small effects, business needs to conclude experiments efficiently, high variability in the metrics of interest etc. This talk will address some of these challenges and highlight the importance of choosing the appropriate statistical method to measure effects reliably.