RFM Segmentation: Recency, Frequency, and Monetary Value in Plain Terms
RFM is the oldest segmentation model in ecommerce, and it still works, because it is built on the three things that actually predict whether a customer will buy again: how recently they bought, how often they buy, and how much they spend.
RFM segmentation scores every customer on three dimensions: recency, how recently they last purchased; frequency, how often they purchase; and monetary value, how much they spend. Each dimension is scored from 1 to 5, usually by quintile, so a customer who bought yesterday, buys weekly, and spends a lot scores 5-5-5, and a customer who bought a year ago, bought once, and spent little scores 1-1-1. The scores combine into segments with names that tell the team what to do: Champions, Loyal Customers, At Risk, and Lost. The model is simple, it runs on data every store already has, and it predicts repeat purchase better than demographics, because it measures behavior, not identity.
How the three scores are built
Recency is the strongest of the three, because a customer who bought yesterday is far more likely to buy again than a customer who bought a year ago, whatever they spent. The recency score is usually built by quintile: the most recent 20 percent of customers score 5, the next 20 percent score 4, and so on down to the oldest 20 percent scoring 1. The lookback window matters: a store that sells weekly should use a shorter window than a store that sells annually, and most RFM models use a 12 to 24 month lookback, because older purchases say little about current behavior.
Frequency is the count of orders in the window, scored the same way. A customer who ordered five times in a year scores higher than a customer who ordered once, and the score should count orders, not items, because a customer who buys five things in one order is not a frequent buyer. They are a big buyer, and that is the monetary dimension's job. Monetary value is usually net revenue after returns, not gross, because a customer who returns half of what they buy is not a high-value customer, whatever the gross says.
The three scores combine, but not as a simple average. Shopify's RFM report scores recency 1 to 5 and combines frequency and monetary into a single FM score, the average of the two rounded down, so frequency 4 and monetary 5 give an FM of 4, and a Champion is recency 5 with FM above 3. A recent big spender who buys rarely is a different segment from a frequent small spender, and the team should treat them differently.
What each segment should get
Champions, recency 5 and FM above 3, are the store's best customers: early access, exclusive products, and a thank-you that is not a discount, because a Champion who gets a discount learns to wait for one. Loyal Customers, high frequency and monetary but slightly lower recency, should get a reason to come back: a new product or a reminder of what they bought last time. Potential Loyalists, recent buyers with low frequency, should get a second-purchase nudge, because the second purchase is the one that turns a buyer into a customer.
At Risk, customers who bought often and spent a lot but have not bought recently, should get a win-back: a personal message, a survey asking what changed, or a product they have not tried. Hibernating, customers who bought once and have not returned, should get a light touch, a single email with the best of what is new, not a discount ladder, because a discount to a customer who bought once and left is a subsidy for leaving. And Lost, customers who have not bought in the longest window and scored low on all three dimensions, should be removed from regular mailings, because emailing them costs deliverability and returns nothing.
The segments are not permanent. A customer moves between them as they buy, or as they stop buying, and the model should be re-run monthly, so the segments reflect current behavior. A Champion who has not bought in six months is not a Champion anymore. They are At Risk, and the team should treat them that way.
What RFM does not do, and what to pair it with
RFM does not explain why a customer behaves the way they do. It says a customer bought recently and often, but it does not say whether they buy for themselves or as a gift, whether they are price-sensitive or brand-loyal, or whether they are about to churn because a competitor launched a better product. That is the limit of a behavioral model, and it is why RFM should be paired with a qualitative layer: a survey, a win-loss interview, or a support ticket review that explains the behavior the scores describe.
RFM also struggles on a store with few customers or a short history. A hundred customers cannot build meaningful quintiles, and three months of trading has no recency distribution to score. For those stores, use absolute thresholds instead: bought in the last 30 days, bought more than twice, spent more than $100.
And RFM is a starting point, not a strategy. The segments tell the team who to talk to and what to say, but not what to sell, charge, or build. The store that runs RFM and sends the same email to every segment has built a model and ignored it. The store that sends each segment a different message, cadence, and offer has used it, and the difference shows up in the repeat purchase rate within a quarter.
Frequently asked questions
What does RFM stand for?
Recency, frequency, and monetary value. Recency is how recently a customer last purchased, frequency is how often they purchase, and monetary value is how much they spend. Each dimension is scored from 1 to 5, usually by quintile, and the scores combine into segments like Champions, Loyal Customers, At Risk, and Lost.
How is the FM score calculated?
Shopify's RFM report combines frequency and monetary into a single FM score, calculated as the average of the two scores rounded down, so a customer with frequency 4 and monetary 5 gets an FM of 4. A Champion is a customer with recency 5 and FM above 3.
How often should RFM segments be re-run?
Monthly. A customer moves between segments as they buy or stop buying, and a Champion who has not bought in six months is not a Champion anymore. They are At Risk, and the team should treat them that way.
What should I send to At Risk customers?
A win-back: a personal message, a survey asking what changed, or a product they have not tried. Do not send a discount ladder, because a discount to a customer who bought often and stopped is a subsidy for leaving.