incrementality

Geographic Experiment design and Evaluation Tool designed to help you determine the true lift of marketing efforts.

This repository provides open-source best practices for for conducting geographic randomized controlled trials (Geo RCTs) for measuring incremental sales effect of advertising cammpaigns. It includes details on one design type in particular, a multi-armed stepped experimental design that has particular advantages in terms of statistical strength.

End-to-end incrementality sandbox on M5-style retail data (DiD + Synthetic Control + placebo tests) with Streamlit dashboard.

PySpark + Hive pipeline measuring campaign incrementality, statistical significance, and ROI on 13M+ user incrementality-test records (Criteo dataset).

Lightweight, transparent marketing mix models for DTC brands. Estimate per-channel causal lift from spend + sales data — with honest diagnostics about when not to trust the result.

Open-source AI marketing measurement & incrementality testing platform. Track every AI creative from prompt to causal revenue lift — A/B experiments, SRM, sequential testing (mSPRT), MMM, Thompson sampling, RLS multi-tenancy. Self-hosted. Built with Claude Fable 5 ultracode.

AI Skill for consistent, reliable 12-hour optimization of marketing campaign budgets across OpenClaw, ChatGPT Codex, and Claude.

Privacy-preserving contextual bandit that optimizes for causal uplift while learning only from differentially-private aggregates — a runnable simulation of causal marketing under the Google Privacy Sandbox, with a real Shared Storage / Private Aggregation browser demo.

Bayesian marketing mix modeling, multi-touch attribution, and causal inference toolkit for unified marketing measurement and incrementality analysis.

Geo-based incrementality testing framework with synthetic control, DiD analysis, and power analysis for measuring true causal marketing lift

Reference implementation of marketing mix modelling, incrementality testing and budget optimisation, validated against known synthetic ground truth.

Causal analysis case study measuring true incremental conversions across paid media audiences using propensity score matching.

Marketing Mix Modeling with Google Meridian on GA4 data. Bayesian inference, full posterior distributions, PyMC-powered. Applied to real GA4 ecommerce data with step-by-step guide.

Paid-media measurement playbook: geo lift, MTA, incrementality, MMM, web attribution. Synthetic data, reproducible.

A repeatable decision framework for evaluating coupon-driven first-purchase experiments

Causal inference platform for marketplace intervention evaluation — DiD, Synthetic Control, PSM, Instrumental Variables, and Event Study with incrementality simulation. Built for Staff-level analytics roles.

Benchmarking geo-lift estimators for ad incrementality: TBR, synthetic control, and TWFE DiD under simulated geo-panel experiments

Synthetic randomized campaign analysis with CUPED and full cost reconciliation; a pre-specified rule returns Redesign when profit uncertainty crosses zero.

Multi-touch attribution graded against a planted ground truth. Six models compete and none recovers it: exact Shapley and a Markov chain both lose to a 40/20/40 heuristic. Settled with a geo holdout. 60 tests.