RFM segmentation bins customers by recency, frequency and monetary value — quantile bins, five by default, with recency inverted — giving each customer an R/F/M triple.
RFM is the classic direct-marketing segmentation. Recency: how recently did the customer last purchase? Frequency: how often have they purchased? Monetary: how much have they spent? Each dimension is binned (1–5, low to high), so each customer ends up in one of 5³ = 125 cells.
Why this works: past behaviour predicts future behaviour in direct marketing surprisingly well — customers who bought recently, often, and at high value are dramatically more likely to respond to the next campaign than the average customer.
Use cases: 5,5,5 (champions) get loyalty perks; 5,1,1 (new low-value customers) get welcome / upsell flows; 1,5,5 (at-risk valuable customers) get retention offers; 1,1,1 (lapsed low-value) often get dropped from the active marketing list.
Recency / Frequency / Monetary numeric columns + optional customer ID + bin count.
Cell-count distribution + top 25 customers by total score. No inferential test.
No p-values here — this is descriptive segmentation. The next step is to test campaign response rates across the cells with a chi-square or A/B test.