RFM analysis: how to calculate it and improve your results

RFM analysis separates three customers with the same revenue this quarter who have completely different futures. One has been buying every month for two years, another came back after a six-month break, a third placed their first order yesterday. In a spreadsheet they look identical. RFM analysis separates these three situations into a single score that can be calculated, compared, and tied to a specific action. Below you will find the formula, the interpretation, and what actually moves the numbers in B2B companies.

What RFM analysis is and how to calculate the score

RFM analysis is a method of evaluating a customer based on three dimensions: recency (how recently they bought), frequency (how often they buy), and monetary (how much money they leave behind). The name recency frequency monetary is simply an abbreviation of these three English words. In B2B, where sales cycles can be long and purchases follow contracts and schedules, these three variables are available in every ERP or CRM system and require no additional research.

The starting point is a transaction table: customer ID, purchase date, amount. From it we calculate three values for each customer:

• R = the number of days from the last purchase to the analysis date. The smaller, the better.
• F = the number of transactions in a chosen period, for example 12 or 24 months. In B2B it is better to count orders than invoice line items.
• M = the total value of purchases in the same period, preferably in margin rather than revenue.

Next, each of the three values is converted into a score, usually from 1 to 5. The simplest approach is to divide into quintiles: customers with the shortest R get 5, those with the longest get 1; the same applies to F and M. A customer's score is a combination of three digits, for example 5-4-5, or a sum from 3 to 15 points. The combination says more than the sum, because it separates an active and frequent customer from one who bought once but spent a lot of money.

Calculating R, F, and M is analysis. Only grouping customers with similar combinations into segments is RFM segmentation, and that is what can be tied to actions.

The added value appears when, instead of manually dividing into quintiles, we use probabilistic models. In the Handbook of Marketing Decision Models (2017), the authors describe the distinction between contractual and non-contractual relationships and the resulting CLV models, in which the probability of a future purchase is predicted at the level of an individual customer. For B2B this means one thing: R, F, and M are only the beginning of a forecast, not its end.

Where a good score comes from, and where a bad one does

An RFM score is not a general judgment of a customer. It assesses their purchasing behavior within a specific time window. Changing the window can reshuffle half the ranking, so before we draw any conclusions, we need to establish what we are measuring.

What raises the score

• A short purchase cycle. A customer who orders quarterly has a better R by definition than one who orders once a year, even if both are loyal.
• Repeatability. Three orders a year mean a higher F than one large order. In B2B, repeatability often means the customer has made us a permanent part of their purchasing process.
• Growing order value. If M grows faster than F, the customer is adding more categories rather than just repeating the same order.
• Activity beyond purchasing. Replies to emails, participation in meetings, downloads of materials. These are signals that, in CLV models, translate into the likelihood of another purchase.

What lowers it

• Long gaps.In B2B, a gap longer than the typical buying cycle signals that the client is being served by someone else or has put the project on hold.
• High-value one-off purchases.High M with low F is often a project, not a relationship. Such a client looks good in the sales report and weak in the forecast.
• Purchases in a single category only.Low M with high F means the client treats us as a supplier of one item, easy to replace.
• Changes in the client's organization.Turnover of decision-makers, mergers, budget changes. None of these appear in transaction data, yet all of them affect R.

Let's return to the distinction from the literature: in non-contractual relationships, typical of B2B, a client does not have to say anything to stop buying. That is why the RFM score should be read together with contact data, not in isolation.

What results are typical in B2B

There is no single benchmark that fits every company. Ranges differ across industries, sales models, and buying cycle lengths. A packaging manufacturer will have different R and F distributions than an industrial software provider, and a regional distributor different ones than a company selling services nationwide.

Instead of looking for a universal number, it is worth establishing your own reference points:

• Median R in your database.If the median is 90 days, a client with R of 200 days is far in the tail, even if in another company they would be a star.
• F distribution.In many B2B companies, more than half of clients have F of 1 or 2. This does not mean they are bad. It means the base is flat and the entire potential sits in a narrow group.
• M concentration.Check what percentage of revenue comes from the top 10% of clients. If it is 60-70%, concentration risk is high and should influence priorities.
• Share of clients in the "champions" segment.In healthy B2B bases this is usually a few to a dozen or so percent. If it comes out at 30%, the thresholds are probably too lenient.

These numbers are not a market benchmark, only a starting point for a conversation about your own base. If you want to compare with what happens in companies in the region, see the material on B2B customer segmentation, where we describe practical differences between industries.

How to improve your RFM analysis score

Improving the score starts with changing the behaviors these scores describe, not with editing the spreadsheet. Below are four actions that in practice translate into movement between segments.

1. Shorten the gap between purchases

The biggest impact on R comes from contact after the purchase, not before it. If a client buys every 90 days, a reminder on day 60 is already late. It is worth setting a contact rhythm for each segment based on order history and keeping to it in your CRM. This action alone raises R, because the client comes back sooner.

2. Increase the number of categories, not the number of orders

A client with F of 4 and M of 4 is often the same product bought four times. A client with F of 4 and M of 5 is four different categories. The second situation is much harder for a competitor to reverse. Instead of pushing for more frequent orders, check which categories the client buys from other suppliers and start with the one that is most complementary.

3. React to a drop in R before the client disappears

In B2B, churn is usually visible in the data before it becomes a fact. If a client has gone past 1.5 times their typical purchase cycle, that is a signal to reach out. This is about checking whether the project has been paused and whether the decision-maker has changed, rather than about sending an offer. It is the cheapest way to keep a client in a high-R segment.

4. Combine RFM with contact data

The RFM score alone tells you what happened. It does not tell you why. Combining it with email response rates, meeting attendance, and support ticket history lets you distinguish a client who buys rarely because that is how their industry works from one who buys rarely because they are unhappy. In our agency practice, we see that these two groups require completely different actions, yet they look identical in RFM alone.

If you want to turn segments into a marketing and sales action plan, see how we build a B2B marketing strategy based on data from your client base. Also keep in mind that RFM segmentation is a tool, not a goal. Its value only appears when someone in the company has an assigned task that follows from a given segment.

Where to start: a plan for the first quarter

1. Pull transactions from the last 24 months out of your ERP or CRM: customer, date, amount, category.
2. Calculate R, F and M for each customer and assign a score of 1 to 5 in each dimension.
3. Build a segment table and check how many customers and what share of revenue falls into each one.
4. Pick two segments that are the priority this quarter, and assign them specific actions plus a person responsible.
5. Set a date for reviewing the results, for example in 90 days, and write down what is supposed to change.

That is enough for RFM analysis to stop being a report and start being a decision-making tool. The rest comes down to consistent execution.