GeoLift Documentation

Welcome to GeoLift – causal inference for geo-experiments using Sparse Synthetic Control (SparseSC).

Overview

GeoLift estimates incremental impact for regional campaigns by comparing treated units to a synthetic control constructed from donor units. The library provides:

  • A pipeline for power analysis, donor evaluation, and inference

  • Recipe scripts and configuration files under data-config/

  • Compact reports and plots for decision support

Key questions

GeoLift helps quantify incremental impact in regional experiments by constructing a synthetic control for treated units. It is often used alongside other measurement approaches (e.g. media mix modelling, attribution, or classic A/B tests) and integrates with standard analytics workflows.

Process overview:

  1. Find fair comparisons (donor evaluation)

  2. Check test strength (power analysis)

  3. Measure the lift (inference and diagnostics)

Documentation Structure

Documentation is organised by task and technical depth:

Getting Started

Technical Reference

Quick Navigation