CLV Prediction: How 12% of Customers Drive 80% Revenue
Industry
Retail - Ecommerce
Services
CLV Prediction · Customer Segmentation · Data Engineering
Timeline
6 weeks
The Situation
In most e-commerce businesses, a small share of customers generates the vast majority of revenue. The problem isn’t that companies don’t suspect this – most do. The problem is that without a systematic way to identify which customers are the most valuable, marketing budgets get spread evenly, retention efforts are reactive rather than targeted and high-value customers get the same treatment as one-time buyers.
We built a CLV prediction system that answers one question: which customers will generate the most revenue in the months ahead and is the answer accurate enough to change how a business allocates budget?
What We Built
We integrated purchase history, website behaviour, acquisition channels, demographics and cart/funnel interactions into a single analytical dataset. For each customer, we calculated 39 behavioural metrics: visit frequency, average order value, recency of last purchase, spending trends, and funnel conversion rates.
On that foundation we built a five-component prediction system: a purchase probability classifier, a segment-level calibration layer, an automatic routing layer that directs each customer to the right spend model and two separate regression models for VIP and standard customers. The output is a specific dollar amount each customer is expected to generate which drives automatic segmentation into five groups: Whale, High-value, Mid-value, Low-value and One-time. Each with a distinct retention and marketing strategy.
The system also flags early warning signals of customer churn before the customer is actually lost based on: increasing days without a visit, declining order frequency and dropping average order value
45% of customers had performed no action in the last six months – effectively already gone. The system detects this trajectory while there’s still time to intervene.
Results
Revenue identified from 25% customers
95
%
Revenue capture improvement
13
%
Distinct Customer Segments
5
Working on a similar problem?
Start with a free iteration - we scope the problem and build the first deliverable at no cost.