Free RFM analysis tool

RFM analysis scores every customer on three things — how recently they bought (recency), how often they buy (frequency), and how much they have spent (monetary) — and groups them by the combination. Each dimension gets a score from 1 to 5, so a customer is summarised by three digits such as 555 or 144. It needs nothing more than a list of orders with a date, a customer id and a value, which is why it is still one of the highest-return analyses in retail marketing.

Work out three numbers per customer: days since their last order, their total number of orders, and their total spend. Then rank customers on each number and split the ranking into five equal groups, giving 5 to the best fifth and 1 to the worst. Recency is reversed — fewer days since the last order is better, so the most recent fifth scores 5. The three digits together form the RFM score, and many teams also use an average of the three as a single overall rank.

Most implementations use around eleven named groups mapped from the recency and frequency-monetary scores. Champions bought recently, buy often and spend the most. Loyal Customers buy often but less recently. Potential Loyalists are recent buyers with a few orders. New Customers bought once, very recently. At Risk and Cannot Lose Them were valuable but have gone quiet. Hibernating and Lost are old, low-frequency and low-value. The names matter less than the fact that each one implies a different action.

There is no absolute good score, because RFM is relative by construction: scores are quintiles of your own customer base, so roughly 20% of your customers score 5 on each dimension no matter how your business is doing. A 555 is your best customer today; a 111 has not bought in a long time, rarely did, and spent little. What you should watch is movement between scores over time, and the share of revenue coming from the top segments.

Customer segmentation is the practice of splitting a customer base into groups that behave differently enough to justify treating them differently. Segments can be built on behaviour (what people bought and when), value (how much they spend), demographics, or lifecycle stage. Behavioural segmentation such as RFM tends to outperform demographic segmentation for marketing decisions, because past purchasing predicts future purchasing far better than age or postcode does.

Monthly suits most retail and ecommerce businesses, quarterly is enough for high-value or slow-cycle categories, and weekly is usually noise. The right cadence follows your purchase cycle: recency scores only move meaningfully once a typical gap between orders has passed. Keep the snapshots, because the value of RFM comes from watching customers move between segments — a Champion sliding to At Risk is the signal, not the label itself.

RFM puts customers into segments you defined in advance using fixed quintile rules, so the groups are consistent, explainable and easy to brief to a marketing team. K-means finds groups from the data itself, without names, and will happily discover that two very different behaviours are sitting inside one RFM label — a high-frequency low-value group and a low-frequency high-value group both scoring the same. Run RFM for the actions and clustering as a check on it.

Match the message to the segment. Champions get early access, referral asks and no discount, because they would have bought anyway. Loyal Customers get cross-sell into categories they have not tried. At Risk and Cannot Lose Them get a genuine win-back with a real incentive, sent while the recency gap is still recoverable. New Customers get onboarding aimed at a second order. Hibernating and Lost get one cheap reactivation attempt, then suppression, which usually saves more than it costs.