Machine Learning & Predictive Analytics

Background:
This is a luxury brand in Asia, with 100k+ members in Hong Kong. Under the fierce competition, the brand asked a bold question: can we identify our most profitable customers as early as possible?

Solution:
Through data discovery and exploratory analysis using 3 years of members and transaction data, we realised that approx. 5% of "big-spenders" contribute to approx. 60% of revenue. With the strategic objective to identify "big-spenders" (i.e. customers who spent a certain amount in a 12 month period), project team is commissioned to build a model to predict whether a first-time customer will be a big spender, in the coming 12 months.

Achievements:
After proper modelling and training, the algorithm is able to correctly predict whether a new customer will be a big-spender, with over 90% accuracy. The algorithm can be deployed in POS, loyalty program, campaign planning, and customer services to assist strategic product/ service recommendation and frontline decision making.