RFM (Recency, Frequency, Monetary) scoring is a proven, interpretable method for segmenting customers and directing marketing effort where it matters most.

Data & Preparation

  • Transactions with customer_id, timestamp, and amount, or a customer table with precomputed recency, frequency, monetary.
  • Choose a time window (6–24 months) that matches your repeat purchase cycle.
  • Handle outliers (e.g., winsorization on spend) and inactive customers per business policy.

Scoring

  • Compute recency as days since last purchase; frequency as number of orders; monetary as total or average spend.
  • Bin each into quantiles (e.g., 1–5) and optionally set weights (e.g., R:2, F:1, M:1).
  • Define segment rules: Champions (R5F5M5), Loyal, Potential Loyalists, At‑risk, Hibernating, New, etc.

Profiles & Actions

  • Profile segment size, value share, repeat rate, AOV, channel mix, and product categories.
  • Playbooks: loyalty rewards (Champions), nurturing (Potential Loyalists), win‑back (At‑risk), onboarding (New).
  • Export CRM lists and track KPIs to measure uplift over time.

Best Practices & Caveats

  • Seasonality: pick a window that covers at least one full seasonal cycle.
  • Heterogeneity: consider segmenting by region or category if behavior differs.
  • Complementary models: pair with CLTV (e.g., BG/NBD + Gamma‑Gamma) and A/B tests for targeting strategies.
Run RFM Segmentation Back to Service Page