Geo Experiment Using Google's Matched Markets Package | Open Source | Time Based Regression

Geo-experimentation is one of the most common ways of carrying out incrementality tests. By now, unless you have been living under a rock, you know that incrementality tests are the gold standard for establishing causality of a marketing initiative.

And among popular approaches for geo-experimentation is the open source package from Google called Matched Markets, which uses the principle of time-based regression (TBR).

The best part about this is that you can run this even if you are not a data scientist - you just need to be a little familiar with Google Colab/Python environments and you can start running it by providing just a couple of inputs.

1. The Cost & Sales data for each geo per day
2. The Geo Eligibility criteria, if any

Later, you also need to provide the post-test analysis data.

At the end of the analysis, you will budget recommendations, split of geos into treatment and control groups and lift analysis.

00:00 What is Time Based Regression?
01:00 Why Matched Markets?
02:13 Why Geo-Experiments in the first place?
04:00 Overview of the Google Colab notebook
04:47 Input Data Sheet: Cost & Sales Data by Geo & Date
05:52 Input Data Sheet: Geo Inclusion Eligibility
08:05 Parameter Selection Process
10:26 Greedy Search for Research Design Selection
16:00 Design Diagnostics & tests
19:15 Post-Test analysis
22:14 Visualization of point-wise and cumulative lift
24:45 Summary

Here is the link to the Google Colab:
https://colab.research.google.com/dri...

Here is the link to the official Matched_Markets Github repo:
https://github.com/google/matched_mar...

Here is the link to the Google Doc with notes:
https://docs.google.com/document/d/1p...

Here are the sheets for the inputs:

https://docs.google.com/spreadsheets/...

https://docs.google.com/spreadsheets/...

Other related videos on incrementality & experimentation:

Incrementality Test Using DiD
   • Incrementality Test to Detect Lift Using D...  

Google CausalImpact vs Meta GeoLift
   • Google CausalImpact vs Meta GeoLift - Comp...  

Statistical Significance vs Duration of Testing - When to stop a test
   • Statistical Significance vs Duration of Te...  

Different ways to measure incrementality
   • Understand the Different Ways to Measure I...  

Incrementality testing with Event Study Model
   • Incrementality Testing with DiD/TWFE/Event...  

BFCM Forecasting
   • Black Friday/Cyber Monday - Forecasting Sa...  

Incrementality Testing for Causal Impact of Google/Meta
   • How to Use Incrementality Testing to Estab...  

Geo-clustering for geoholdout test
   • Geo Clustering for Geo Holdout Incremental...  

Measuring incrementality of branded search
   • Measuring Incrementality of Branded Search...  

PMax incrementality testing template:
   • Performance Max Incrementality Test | How ...  

Multi-channel incrementality testing template:
   • Incrementality Lift Template - Multi-Chann...  

Understanding geo-experiments | Geo-Holdouts | Causal Inference
   • Understanding Geo Experiments in Marketing...  

#statisticalanalysis #incrementality #geoexperiment #marketinganalytics #marketingmeasurement #marketingeffectiveness #marketingscience #causalinference