RFM Analysis: What It Is and How to Score Customers

By Alva Chew · Updated 2026-11-05

Definition RFM analysis ranks each customer on three numbers: how recently they bought (recency), how often (frequency) and how much they spend per order (monetary value). Score each from 1 to 5, add the scores, and the top group is your best customers while the bottom group has drifted away. It needs sales tied to named customers, which a points program provides.

What is RFM analysis?

RFM analysis sorts customers by three facts about their buying. Recency is how long ago they last bought. Frequency is how many times they bought in a period. Monetary value is how much they spend per order. Together they show who your best customers are and who has gone quiet.

In their 2005 paper, Peter Fader, Bruce Hardie and Ka Lok Lee note that in direct marketing it is common practice to summarise a customer’s past behaviour with these three measures. They define monetary value as the average purchase amount per transaction.

How do you calculate RFM scores?

Pull one row per customer: date of last purchase, number of orders in the period, and average order value.
Rank customers on each column.
Split each ranking into five equal groups. Give the best group a 5 and the worst a 1. For recency, the most recent buyers get the 5.
Add the three scores, or keep them as a three-digit code such as 5-4-2.

Five groups is a common convention, not a rule. Fewer groups suit a small customer list.

What does an RFM table look like?

These members are invented to show the arithmetic. The period is the last 12 months and today is 5 November.

Member	Last purchase	Orders	Average order	R	F	M	Total
A	2 days ago	38	US$9	5	5	3	13
B	9 days ago	14	US$22	5	4	5	14
C	40 days ago	20	US$8	3	5	2	10
D	160 days ago	11	US$18	2	3	4	9
E	290 days ago	2	US$7	1	1	1	3

Member B has the highest total even though A visits more often, because B orders more each time. Member D bought 11 times but not for five months, which is a warning sign a total alone would hide. Always read the three scores, not just the sum.

What do you do with the groups?

High on all three (13 to 15): your core customers. Make sure redeeming points at the counter is easy for them.
High frequency, low monetary: regulars with small baskets. See average order value for ways to lift spend.
High monetary, falling recency: big spenders who have slipped. These are the first people to contact when you win back lapsed customers.
Low on all three: one-off visitors. A small, simple points reward may bring some back, but do not spend heavily on them.

What are the limits of RFM?

RFM looks backward. Fader, Hardie and Lee studied 23,570 customers who first bought from the online retailer CDNOW in early 1997. They found that recency and frequency each had a positive link to later spend. They also found that the grid of observed data was sparse and “somewhat untrustworthy”, because some recency and frequency combinations had no customers at all.

Finding	Detail
Customers studied	23,570 first-time CDNOW buyers, Q1 1997
Data window	78 weeks, split into two 39-week periods
Result	Recency and frequency each positively linked to later spend
Caveat	Observed grid sparse, some combinations empty

Source: Fader, Hardie and Lee, “RFM and CLV: Using Iso-Value Curves for Customer Base Analysis”, Journal of Marketing Research 42(4), 2005. One online music retailer, not a café or shop.

Two practical limits follow. Small customer lists produce thin groups, so treat scores as a guide. And RFM does not say why someone stopped coming. A quiet member may have moved away rather than lost interest. For a forward-looking figure, see customer lifetime value.

What data do you need for RFM?

Every purchase must be linked to a person. A till records sales, not people, so cash buyers who never identify themselves drop out of the analysis. The result then describes your identified customers only, who are usually your keenest. Say so when you report it, and compare it with your overall repeat purchase rate and purchase frequency.

A points-based loyalty program supplies that link. Each member has an account, and points are earned in proportion to what they spend, so the spend you record per member matches what you would score in the M column.

Running RFM with Gimify

Gimify members join by scanning a QR code and entering an email, and each member holds one points balance across every outlet. The free plan covers one outlet with unlimited members, so you can build a named customer list before you decide how to analyse it.

Questions owners ask

What does RFM stand for?

Recency, frequency and monetary value. Recency is the time since a customer's last purchase, frequency is the number of purchases in a period, and monetary value is the average amount spent per purchase.

How do I do an RFM analysis without special software?

List each customer with their last purchase date, number of orders and average order value. Rank each column, give scores from 1 to 5, and add them. A spreadsheet is enough for a few thousand customers.

How many customers do I need for RFM to work?

There is no fixed minimum. With very few customers the groups are too small to tell apart, so start by just sorting members by last visit and by orders, then add scoring as the list grows.

Is RFM the same as customer lifetime value?

No. RFM describes past behaviour. Customer lifetime value estimates future spend. Researchers have built models that turn RFM inputs into a lifetime value estimate, but a simple score does not do that by itself.

Can I run RFM if customers pay in cash and don't give a name?

Not reliably. RFM needs every purchase linked to the same person. Anonymous sales cannot be grouped by customer, so the analysis only covers customers who identify themselves, such as loyalty members.