Problem
Revenue stagnation despite growing acquisition spend. High one-time purchase rate with low customer lifetime value across all cohorts. The business had no visibility into which customers were worth investing in.
Data
12 months of transactional data across 8,400 customers. SQL queries on order history, return rates, and session frequency. Data cleaned and structured via Python pandas pipelines.
Solution
RFM segmentation (Recency, Frequency, Monetary) applied in Python. K-means clustering confirmed four actionable customer tiers. Results visualised in an interactive Tableau dashboard.
Recommendation
Priority retention programme for the "at-risk champions" cluster, representing 23% of customers but 41% of revenue. Personalised email sequences and tiered loyalty incentives proposed by segment.
Impact
Projected 18% increase in 6-month CLV for the reactivated segment. Predicted 34% reduction in churn for the high-value cohort, validated at 95% confidence with a holdout test group.
Method
K-means clustering for segmentation, logistic regression for churn prediction, cohort analysis for retention curves. Every result statistically validated before recommendation.