Katkat
Back

Design A/B Test to Isolate Product Usage Drop Causes

GoogleData Scientist · Technical Screen
MediumData ScientistTechnical ScreenAnalytics & Experimentation

Last updated

Evaluates experimentation strategy for diagnosing a simultaneous product usage drop in the U.S. and Mexico. Strong answers identify confounders, design tests or quasi-experiments, control variables, and report causal lift.

Scenario

Product usage dropped 10 % in the U.S. and 11 % in Mexico.

Question

Identify potential confounders, design an A/B test to isolate the cause, specify the variables you would control, and explain how you would report findings to stakeholders.

Hints

Segment users, hold out controls, present causal lift estimates.