Key takeaways
- Averages in reporting hide how customer groups are changing.
- RFM scores customers by recency, frequency and monetary value – straight from order data.
- The real value comes from repeating the analysis and watching customers move between segments.
Contents
Many online shops treat their customers as fixed groups: regulars, new customers, perhaps a VIP segment. Yet customer groups change all the time. Someone who orders regularly today can be gone in six months – without it showing up in monthly revenue right away.
The problem: most reports show totals. Revenue, orders, average basket size. These figures can look stable while buying behavior underneath shifts considerably.
This is not a niche issue. Customers expect relevance: in a US survey, 71 % of consumers expected personalized interactions, and 76 % were frustrated when that didn’t happen (McKinsey, 2021). A shop that doesn’t know its segments wastes marketing budget and misses revenue.
Why averages are misleading
Flat revenue can hide a lot:
- Good customers order less often, and new customers make up for it in the short term.
- New customers come back for a second purchase less often than they used to.
- A loyal segment drifts to a competitor, slowly and quietly.
If marketing budgets keep following old segments, campaigns miss their target. Money goes to groups that have long lost their value.
Metrics show results. Behavior shows change – and it shows it earlier.
What RFM analysis measures
RFM analysis rates customers on three simple questions. All three can be answered directly from a shop’s order data.
- Recency: When did the customer last buy? The more recent, the more active the relationship.
- Frequency: How often did they buy within a period? Frequent purchases point to loyalty and satisfaction.
- Monetary value: How much have they spent in total? Note: a high value alone is not a good sign if the last purchase was a long time ago.
Each dimension gets a score, for example from 1 to 5. A customer scored 5-5-5 bought recently, often and a lot. A customer scored 1-1-1 is practically lost. The combinations form segments that are far more precise than age, region or revenue band.
How the RFM score is calculated
A common RFM scoring method works in three steps:
- Compute three values per customer: days since the last order, number of orders in the period (for example the last 12 or 24 months) and total revenue in the same period.
- Rank and split: sort all customers by each value and divide them into equal groups – five groups (quintiles) give scores from 1 to 5. For recency the order is reversed: the most recent buyers get a 5.
- Combine: each customer ends up with a three-digit RFM score such as 5-4-3. These scores are then grouped into a handful of named segments.
Equal groups have one advantage: the scores adapt to each shop’s own buying rhythm. A shop selling coffee and a shop selling furniture get comparable segments without fixed thresholds.
Typical RFM segments and what to do with them
Four segments, four tasks
R high · F high · M highStrengthen loyalty
R falling · F high · M highWin back personally
R high · F lowLead to a second purchase
R high · F high · M lowGrow the basket
Keep top customers
Customers who buy recently, often and a lot are the most valuable group. The focus here is loyalty, not discounts: early access to new products, personal recommendations, particularly good service. Losing one of these customers hurts.
Win back at-risk regulars
Customers who used to buy often and a lot, but haven’t for a while, are the most important early warning signal. In standard reports they disappear first. A personal approach, a relevant offer based on past purchases or a short question about the reason often works better here than any acquisition campaign.
Lead new customers to a second purchase
New customers have high recency but no frequency yet. The second purchase is the decisive step. Good welcome flows, recommendations that fit the first purchase and a clear reason to come back lay the groundwork.
Develop potential
Customers who buy often and regularly, but spend small amounts, have potential. Complementary products, bundles or subscriptions can raise their value without having to win new customers.
Use budgets more precisely
RFM segmentation helps to allocate marketing budgets more precisely. Instead of broad campaigns, money flows to the segments with the greatest potential for growth or win-back. At the same time, spending on barely profitable groups becomes visible. According to McKinsey, personalization typically drives a 10 to 15 % revenue lift, ranging from 5 to 25 % depending on industry and execution (McKinsey, 2021).
The real benefit: making movement visible
RFM analysis shows its full strength not as a one-off report. Repeated monthly or quarterly, it shows how customers move between segments.
- Churn: if recency and frequency drop among previously strong customers, churn has begun. It shows up here much earlier than in revenue.
- Segment shifts: if many customers move from “loyal” to “at risk” at the same time, there is usually a bigger problem: with the product, the service, prices or delivery times.
- Potential: customers with high recency and frequency but low value are candidates for targeted cross-selling.
This turns an analysis into an early warning system. The reaction no longer comes when revenue falls, but when behavior changes.
A thought experiment: what RFM reveals
Imagine a shop for personal care products comparing its RFM segments quarter by quarter. Total revenue is stable. But the analysis shows that a noticeable share of former top customers has slipped into the “at risk” segment. Their last purchase was much longer ago than usual. At the same time, a promotion brought in many new customers who offset the revenue.
In the monthly report, everything looks fine. The RFM view shows a clear risk instead: the most valuable customers are leaving. The shop starts a personal approach for exactly this group and checks in parallel whether delivery times or range changes are the cause. Without RFM, this would only have surfaced once revenue dropped.
A simplified example for illustration.
A pragmatic way to start
No big project is needed. An export of orders with customer ID, date and amount is enough to begin with.
- Start simple: use three levels per dimension instead of five. That gives 27 groups and stays manageable.
- Name a few segments: combine the groups into four to six clear segments, such as top customers, at risk, new customers, potential and lost.
- Assign actions: define one concrete action per segment, for example in the email marketing tool or in campaign management.
- Repeat regularly: only the comparison over time reveals the important changes.
Conclusion
RFM analysis is simple, quick to set up and surprisingly revealing. It shows which customers are gaining value, which are stagnating and which are leaving right now. Looking at customers this way makes marketing and service more targeted – and makes it possible to react to changes before they show up in revenue.
Frequently asked questions
What does RFM stand for?
RFM stands for recency, frequency and monetary value – how recently, how often and how much a customer buys. Together, the three values describe a customer’s buying behavior.
What is RFM segmentation?
RFM segmentation groups customers by their RFM scores into segments such as top customers, at risk, new customers or lost. Each segment gets its own marketing and service approach.
What data is needed for an RFM analysis?
An export of orders with customer ID, order date and amount. Every online shop has this data.
How often should an RFM analysis be repeated?
Monthly or quarterly, depending on how often customers buy. Only the comparison over time shows how customers move between segments.
Does RFM work for B2B shops?
Yes. Regular orders are typical in B2B. Deviations from the usual rhythm are a particularly clear warning sign there.