Customer Segmentation

Unsupervised · Live

Auto-Computed Features

Recency is calculated automatically from the date you enter. You don't need to compute days yourself, just pick the date of the last purchase.

Fields filled

0 / 2

What the model evaluates

Last order date → recency in days since purchase

Total lifetime spend → monetary value (USD)

Log transform applied to both before clustering

Frequency dropped — 90%+ of customers had only 1 order

K=3 selected via Elbow + Silhouette (score 0.4129)

Possible segments

Active Customers

Purchased recently (~40 days avg). Nurture and upsell.

Lapsed High-Value

High historical spend (~$138), inactive ~278 days. Win-back candidate.

Lost Cheap

Low spend (~$38), inactive ~281 days. Low reactivation priority.

Trained on the Olist Brazilian E-Commerce dataset (83,246 customers). Segmentation reflects behavioural patterns from 2016–2018 transaction data. Results are indicative, not prescriptive.