RFM analysis is a way of grouping customers using three facts from their order history: how recently they last bought (recency), how often they buy (frequency) and how much they spend (monetary value). Scoring every customer on those three measures shows who your best customers are, who is drifting away and who bought once and never came back.
How RFM analysis works
You start with an export of orders: a customer identifier, the order date and the order value. For each customer you work out three numbers: days since their last order, how many orders they placed in the period you are looking at, and their total spend in that period. Choose a look-back period that fits how often people normally buy: 12 or 24 months suits many shops, while fast-repeat products need less.
Each number is then turned into a score, usually from 1 to 5. A common method is to rank customers on each measure and split them into five equal groups, so the most recent fifth gets a recency score of 5 and the least recent fifth gets 1. A customer scored 5-5-5 bought lately, buys often and spends well. A customer scored 1-1-1 bought once, a long time ago, and spent little.
The scores are then gathered into named segments that you can act on. Typical examples:
- Best customers High on all three. Look after them, and avoid training them to wait for discounts.
- New customers High recency, low frequency. The second order is the one to win.
- At risk Used to buy often and spend well, but have not ordered for a while.
- Lapsed Low recency and low frequency. Worth one or two attempts, rarely worth chasing hard.
A spreadsheet handles this comfortably for a few thousand customers. Several ecommerce and email platforms now offer RFM-style groupings built in, so check yours before building one by hand.
Why it matters
Most online shops treat their customer list as one audience and send everyone the same email with the same offer. RFM analysis shows how uneven that list really is. A small group of repeat buyers often accounts for a large share of revenue, and they need a different message from someone who bought one item in a sale two years ago.
It also ties marketing to customer lifetime value. Once you can see which first purchases tend to lead to repeat buyers, you can judge what it is worth paying to win a customer, rather than judging every campaign on its first order alone.
For UK businesses there is a data protection side. Scoring customers this way is profiling under UK GDPR. For ordinary marketing segmentation most retailers rely on legitimate interests, but your privacy notice should say that you analyse purchase history to personalise marketing. The emails you then send to each segment still need a lawful basis under PECR, such as consent or the soft opt-in for existing customers.
Common mistakes
- Scoring once and never refreshing. Recency changes every day, so rerun the analysis monthly or automate it.
- Using a time window that does not suit the product. A mattress shop and a coffee subscription have very different normal gaps between orders.
- Counting refunded and cancelled orders in the monetary score, which flatters customers who send most things back.
- Treating guest checkouts under different email addresses as different people, which splits one loyal customer into several weak ones.
- Building careful segments and then sending all of them the same discount.
How to act on it
Start small. Export 12 to 24 months of orders, score customers in a spreadsheet and look at how revenue splits across the segments. That first view alone often changes how a business talks about its customers.
Then give each segment one job. New customers get a strong post-purchase email sequence aimed at the second order. At-risk customers get a reminder of what they bought before, or what is new in that range, before any discount is offered. Your best customers can seed a lookalike audience for paid social and be excluded from acquisition campaigns that would otherwise pay to reach people who already buy from you. Lapsed customers get a final attempt or two; if they have also stopped opening your emails, move them under a sunset policy so they stop dragging down your email deliverability.
Measure results by segment: the share of new customers who place a second order, and how many at-risk customers come back. Turning a customer list into a retention and acquisition plan is part of my digital marketing strategy work.