Estimating the causal lift of a price promotion

A case-study template for measuring whether a promotion increased incremental revenue rather than merely shifting existing demand.

Decision Question

Did a temporary price promotion create incremental revenue, or did it mostly pull forward purchases that would have happened anyway?

Why This Is Causal

A predictive model can forecast sales during promotions, but the decision requires a counterfactual: what would revenue have been without the promotion?

Identification Strategy

Start with a difference-in-differences design comparing treated products or markets to a credible comparison group before and after the promotion.

Key assumptions to address:

• Parallel trends before the promotion.
• No simultaneous shocks that differentially affected treated units.
• Stable product definitions and comparable measurement windows.
• Limited spillovers across products or markets.

Notebook Plan

Add notebooks such as:

• notebooks/01-data-audit.ipynb
• notebooks/02-parallel-trends.ipynb
• notebooks/03-did-estimation.ipynb
• notebooks/04-sensitivity-analysis.ipynb

Executive Summary Template

Replace this with the final business-facing takeaway:

The estimated effect was X with uncertainty interval Y. The result is most credible for Z segment. The recommended decision is A, with B caveat.