How Can You Use RFM Customer Analysis to Increase Repeat Sales?
A store's customer database may contain several thousand contacts. That looks impressive. But the number itself says almost nothing about the quality of that database.
One customer orders every month and gradually increases their spending. Another arrived during a sale, bought one inexpensive item, and never returned. A third used to be a regular customer and then suddenly disappeared. Perhaps they switched to a competitor. Perhaps they simply forgot about the store.
Sending all three the same mailing is not the best idea. This is where RFM analysis comes in. It shows who really matters to the business right now, who can still be brought back, and which customers no longer warrant most of the marketing budget.
What Does RFM Analysis Show?
The method is based on just three metrics:
- R — Recency: how much time has passed since the last purchase;
- F — Frequency: how often the customer placed orders;
- M — Monetary: how much money they spent over the selected period.
Suppose two people each spent $500 at the store. But one placed ten orders and was active last week, while the other spent the entire amount a year ago and has not appeared since. In terms of total revenue, they look the same. In terms of RFM, they do not.
That is the point of RFM analysis of a customer database: it helps reveal individual customers' behavior behind the overall statistics.
The method is used by online stores, coffee shops, salons, restaurants, and other businesses with repeat purchases. The intervals, however, will differ. For a coffee shop, a customer who has not visited for a month may already be considered at risk of leaving. For a furniture store, that interval is perfectly normal.
Why Divide Customers into Groups?
When the database is small, a business owner may feel they already know their customers well. Over time, this stops working. The number of customers increases, order histories grow, and spotting changes manually becomes difficult.
RFM analysis in marketing helps answer very practical questions:
- who generates most of the revenue;
- which regular customers have started buying less often;
- who should be offered higher cashback;
- who can be brought back with a reminder about rewards;
- which segments are not worth spending money on expensive advertising.
After this segmentation, mailings become more targeted. A regular customer does not need a “We miss you” message if they visited the store yesterday. And a customer who has not returned for eight months is unlikely to respond to a routine announcement about new stock.
How Do You Calculate RFM?
First, select a period — for example, the past six months or a year. Then collect three values for each customer.
| Metric | What to Calculate |
|---|---|
| Recency | How many days have passed since the last order |
| Frequency | How many purchases the customer made during the selected period |
| Monetary | The total amount they spent |
Next, assign a score to each value. A scale from 1 to 5 is often used: one indicates a weak result, and five a strong one.
The business sets its own thresholds. There is no ready-made scale that works equally well for everyone. It makes no sense for a clothing store and a coffee shop to assess purchase frequency using the same rules.
Let's look at a hypothetical example.
A store analyzes orders from the past year and sees the following picture:
| Customer | Last Order | Number of Purchases | Amount | RFM Code |
|---|---|---|---|---|
| Oleksii | 8 days ago | 14 | $820 | 555 |
| Mariia | 57 days ago | 6 | $260 | 333 |
| Serhii | 230 days ago | 2 | $70 | 111 |
Oleksii purchased recently, bought frequently, and spent more than the others. He receives the highest scores on all three parameters.
Mariia has average figures. She is not lost yet, but her activity is already worth monitoring.
Serhii has not returned for a long time, has placed few orders, and has spent little. Spending the same advertising budget on him as on Oleksii is hardly sensible.
This is how the classic RFM analysis method works. Each customer ends up with a three-digit code, after which similar codes are combined into larger segments.
Three Groups That Are Easy to Work With
In theory, there can be many segments: new, promising, regular, loyal, becoming inactive, lost, and so on. But there is no need to start by building a complex system with ten or fifteen categories.
Three main groups are often enough.
Best Customers
They purchased recently, return regularly, and generate a significant share of revenue. These people already know the company and are most likely satisfied with their previous experience. Still, their loyalty should not be taken for granted forever.
It is better not to overwhelm regular customers with ordinary discounts. Early access to new products, a higher loyalty program tier, additional rewards, or exclusive offers tend to be more appealing to them.
The aim here is not to sell something else urgently. It is more important to maintain the habit of returning and show that their consistency is noticed.
Customers Who Can Still Be Won Back
This is perhaps the most interesting segment. These people used to buy quite actively, but recently have started appearing less often or stopped coming altogether.
The reasons may vary. A customer may have forgotten about the company, given up waiting for a particular product, been disappointed with their last order, or simply received a more convenient offer from a competitor.
A separate approach can be prepared for this group:
- remind them about accumulated rewards;
- offer higher cashback on the next purchase;
- show products from a familiar category;
- set a short validity period for a personalized offer.
It is better to build the message around the customer's previous experience. If they have always bought skincare products, an offer from a random category is unlikely to bring them back any faster.
Lost Customers
They have not bought anything for a long time, barely respond to messages, and usually have a low Recency score. Some of these people can be brought back, but expectations should remain realistic.
Continually increasing the discount is risky here. At some point, the customer will return only for a very low price and disappear again after the order.
It is more sensible to run a separate campaign: offer a gift for returning, award a limited bonus, or ask why the person stopped buying. The last option is sometimes more useful than another promotion. At least it reveals the reason for leaving.
If the segment does not respond at all, it is better to reduce message frequency. The marketing budget is not unlimited.
How Do You Choose Scoring Thresholds?
One common mistake is to take ready-made intervals from someone else's article and apply them unchanged.
Imagine that everyone who purchased in the last 30 days receives a Recency score of 5. For a jewelry store, that may be too strict: people rarely order jewelry every month. For a coffee shop, on the other hand, it is too lenient.
It is better to base thresholds on your own data. Look at how often your customers usually return, what amounts are considered high, and how many orders really distinguish a regular customer from an occasional one.
Sometimes it is convenient to divide the database into equal parts. For example, the 20% of customers with the most recent purchases receive R5, the next 20% receive R4, and so on. In another business, fixed intervals may be clearer: up to 14 days, 15 to 30, and 31 to 60.
Both approaches are valid. What matters is that the scores reflect the audience's actual behavior, rather than simply looking good in a spreadsheet.
How Does ABC/RFM Analysis Differ from Standard RFM?
Classic RFM considers the time of the last purchase, order frequency, and the amount spent. ABC/RFM analysis adds another layer — grouping customers by their contribution to revenue or profit.
Three classes are usually distinguished:
- A — customers with the greatest contribution;
- B — customers with average value;
- C — customers with a small contribution.
Then the R, F, and M metrics are analyzed within each class.
This helps reveal differences that ordinary segmentation sometimes smooths over. For example, two customers may have similar RFM codes, but one buys high-margin products while the other buys only promotional items. Their activity is formally similar, but their value to the business differs.
For a small database, this level of detail may be unnecessary. But with a wide product range and thousands of customers, it gives a more accurate picture.
How Does a Loyalty Program Simplify Analysis?
Calculating RFM manually is possible. You need an order history, a spreadsheet, and a little time. The problem starts later: the data keeps changing.
A customer places a new order — their Recency is updated. Makes two more purchases — Frequency changes. Spends a large amount — Monetary increases. In a large database, recalculating all of this manually every week is inconvenient.
A loyalty program handles the routine work. It connects purchases to individual customers, stores transaction amounts, calculates visit frequency, and updates segments.
The business can then set up separate actions for each group.
For example, a top customer gets access to new products. An at-risk customer receives a reminder about rewards. A lost customer is sent a single personalized offer — without an endless series of identical mailings.
In this case, RFM analysis of the customer database stops being a spreadsheet opened once a year. It becomes a working tool: it helps select audiences for campaigns, avoid giving discounts to everyone indiscriminately, and spot customers who are starting to leave in time.
Analysis itself does not increase sales. It only shows where to look for opportunities. Growth begins later — when the business changes its offers, tests different approaches, and checks which groups have actually started returning more often.
Frequently asked questions
RFM analysis is a method of segmenting customers using three metrics: Recency — time since the last purchase, Frequency — how often they buy, and Monetary — the amount spent over a selected period.
For each customer, calculate the days since the last order, the number of purchases, and the total amount spent during the selected period. Assign each metric a score, often from 1 to 5, and combine them into a three-digit RFM code.
Three groups are enough to start with: best customers who buy regularly; customers who can still be won back after becoming less active; and lost customers who have not purchased for a long time and barely respond to messages.
Set thresholds using your own data on how often customers return, their order counts, and purchase amounts. You can divide the database into equal groups or use fixed intervals. There is no universal scale for every business.
A loyalty program connects purchases to individual customers, stores transaction amounts, calculates visit frequency, and updates segments. This helps businesses set up separate offers and reminders for each group.
