top of page
Data Analyst

AI/ML-DRIVEN CUSTOMER CHURN REDUCTION

Global eCommerce and Marketing Platform
Customer Retention | AI Program Management | Machine Learning

Challenge

 

A major department store’s BOPIS (Buy Online Pickup In Store) offering accounting for 38% of orders broke during the peak period leaving customers without their holiday gifts, damaging the brand. 

 

Approach

 

Provided supply chain & retail subject matter experts to conduct a process analysis of the interconnected retail BOPIS process at a leading department store. Identified key themes and opportunities to improve the end-to-end process, spanning 10+ areas of the business and overall customer experience. Developed a 3-5 year roadmap prioritizing the opportunities and an initial 1 year implementation plan focused on speed to market.

 

Value Delivered

 

•Identified 50 short-, medium-, and long-term opportunities requiring both process and system changes to protect the exponentially growing BOPIS revenue stream of ~$1B.

•After executing the prioritized recommendations, the major department store realized a successful holiday peak season the following year including 45% growth in BOPIS sales.

CHALLENGE

The client built an early machine learning model to predict customer churn, but the organization didn’t trust the model’s accuracy and didn’t understand how to use it.  

APPROACH

Our team developed an operating model for implementation and improvement of the ML model. We designed strategies for using and adopting the churn prediction model across multiple teams, customer segments, and worked with data science and analytics teams to build understanding and interest with business stakeholders. After prioritizing use cases based on prospective business value, we structured the collaborative feature engineering process to improve model actionability and delivered high-risk customers to teams where they could engage with retention tactics.

VALUE DELIVERED

Improved customer churn prediction model accuracy and actionability.

Implemented the model outputs into the client’s production system, enabling stakeholder use for things like dynamic targeting of customer success outreach.

bottom of page