← Back to case studies
Fashion Retail
Fashion Retail: Personalized Win-Back
Value
23.3% less churn
Timeline
6 weeks
The Problem
Lack of a churn model and intransparency into tracking data and online footprints of customers turned reactivation campaigns into guesswork instead of challengable and measurable activities.
The Solution
Processing pipeline combined online and offline data and connected to customers. Look-alike models filled gaps between customer profiles. Churn models leveraged previously implemented segments to deliver custom churn estimates.
Tech Stack
PythonPostgreSQLPower BISageMaker
The Impact
13
13x more customer reach with look-alikes
87%
87% of customers received optimal channel
23.3%
23.3% less churn