Evaluate Campaign Lift with Predictive Analytics and Validation Strategy
Overview
This question evaluates a data scientist's competency in causal inference and uplift estimation, time-series feature engineering (seasonality, adstock/carryover), KPI definition for incremental units/revenue/margin, and validation and diagnostic strategies for measuring marketing campaign lift.
Boston Consulting Group
Aug 4, 2025, 10:55 AM
Data Scientist
Technical Screen
Analytics & Experimentation
44
0
Evaluate Marketing Campaign Lift (Weekly SKU-Level, 3 Years)
Context
You have 3 years of panel data at weekly SKU (and optionally region/store) granularity for a national retailer. The client runs weekly SKU-level marketing campaigns (e.g., spend, impressions, channels, creative) and wants to estimate causal lift from these campaigns.
Task
Design an analytical approach to quantify campaign lift and translate findings into actionable guidance for future campaigns. Clearly specify:
- Target variable (Y) and KPI(s)
- Predictors (X) and key feature engineering (including seasonality)
- Chosen causal/modeling approach(es)
- Validation and diagnostic strategy
- Key caveats/assumptions
- How the insights will inform future campaign planning and targeting
Consider
- Causal inference options: difference-in-differences (including staggered adoption), uplift modeling (treatment effect heterogeneity), regression with controls / doubly robust learners
- KPI definition (incremental units, revenue, margin, ROI/pROAS)
- Seasonality, trends, holidays, and adstock/carryover
- Data pitfalls: selection bias, overlap, stockouts, cannibalization, interference
Solution
Login required
Show