RFM Customer Segmentation Score

RFM: customer segmentation by Recency, Frequency, Monetary

RFM is a behavioral segmentation framework that scores each customer on three things: Recency (days since the last purchase), Frequency (how many orders) and Monetary (total spend). You split each one into quintiles (1–5), and every customer ends up with an RFM = X-Y-Z code. There are fancier statistical models out there, like the NBD/BG-NBD work by Bruce Hardie and Peter Fader (Wharton), which treat purchase behavior probabilistically. Even so, plain RFM is the one most teams actually reach for. Take a concrete example: someone who bought 10 days ago (R=5), 18 times in 12 months (F=5) and spent R$ 4,200 (M=5) lands as 555 — Champion. A customer at R=1 (180+ days), F=4, M=4 is At Risk — they used to be great and they're slipping. R=5, F=1 means New. And 1-1-1 is Lost.

Applications: CRM, marketing decisions, e-commerce

Where RFM really earns its keep is in marketing budget allocation. Champions (555, 554, 545) are worth VIP perks and upsell, since they'll convert on almost any offer. The Loyal group (455, 454) is your audience for cross-sell and referral programs. At Risk customers (high F and M but low R) call for win-back campaigns, maybe a discount or a personalized email. New customers (5-low-low) want a proper onboarding flow. As for the Lost bucket (111, 112), pouring more money in usually doesn't pay off, because the ROI runs negative. You'll see this in e-commerce, retail, SaaS churn analysis, really anywhere people buy more than once. HubSpot, Klaviyo and Salesforce all ship RFM segmentation out of the box.

FAQ

Why quintiles instead of fixed thresholds? Quintiles move with your customer base. The top 20% by spend is always the top 20%, no matter the size of the business or the time of year. A fixed cutoff (say, R$ 1,000 = high) falls apart the moment the market shifts.

Which dimension matters most? Most of the time it's Recency. A customer who hasn't bought in a year probably won't convert, however much they once spent. Some businesses weight things differently, though; in subscriptions, for instance, Frequency tends to dominate.

Does RFM work for B2B? It does, with a couple of tweaks. The windows get longer (Recency in months or quarters rather than days), and Monetary often becomes margin or contract value instead of raw spend.

RFM vs. cohort analysis? Think of RFM as a snapshot of who your customers are right now, while cohort analysis follows how groups evolve over time. You'd use them together, not one in place of the other.