Marketing ScienceADA Asia2022 – 2024Production

Campaign Performance Prediction Engine

Two-stage system predicting campaign pipeline potential before launch and monitoring active campaigns daily.

Daily

Campaign Scoring

2-Stage

Architecture

The Problem

Marketing teams were reactive — they only knew a campaign underperformed after it ended. They needed forward-looking predictions to decide which campaigns to scale, which to cut, and how to reallocate budget.

What Was Built

Two-stage architecture: (1) XGBoost Classifier predicting high vs. low potential campaigns (pipeline threshold classification). (2) XGBoost Regressor estimating expected pipeline value/volume for predicted high-potential campaigns. Features include campaign attributes, budget, type, date-derived features, pipeline context at campaign start, active campaign counts, and historical averages for similar campaigns.

Business Impact

Enabled proactive campaign management — predicting success before launch and flagging underperformers for budget reallocation, shifting marketing teams from post-campaign hindsight to forward-looking intelligence.

Tech Stack

PythonXGBoostscikit-learnBigQuery

Domain Tags

Campaign PredictionMarketing IntelligenceBudget OptimizationMarketing Science

Details

Role
Primary Owner
Status
Production
Tier
Tier 1
Period
2022 – 2024
Employment
ADA Asia