Back to projects
PythonStrategySQLData

A/B Testing Framework

Built an end-to-end experimentation framework for the marketing team, including statistical analysis pipelines.

24Experiments launched
34%Conversion lift
$150KAd spend saved
Project visual

Overview

A reusable framework that allows the marketing team to design, run, and analyze A/B tests without engineering support.

The Problem

The marketing team was running experiments ad-hoc with no standardized process, leading to inconclusive results and wasted budget.

Approach

Created templates for experiment design, built Python notebooks for statistical analysis, and established governance around sample sizes and significance thresholds.

Tools Used

PythonJupyterSQLGoogle AnalyticsStatsmodels

Outcomes

  • Launched 24 experiments in first quarter
  • Improved email conversion rate by 34%
  • Saved $150K in ad spend through better targeting

Learnings

Frameworks are only useful if they are accessible. Documentation and training were as important as the technical implementation.