What Is RFM Analysis: Ranking Ecommerce Customers by 3 Metrics to Decide Revenue Tactics

"We're getting more new customers. But honestly, I can't tell who our most valuable customers are, or who's about to leave." That's a common state in ecommerce. The standard way to put customers in order of importance is RFM analysis.

RFM analysis ranks customers by three numbers so you can decide who gets which tactic. This article covers what the three metrics mean, how to segment customers, what to do for loyal and at-risk customers, and how RFM differs from the cohort analysis it's often mentioned alongside — with charts and examples.

Key Takeaways#

1. RFM analysis = ranking customers by recency, frequency, and monetary value

   The name comes from Recency (last purchase date), Frequency (number of purchases), and Monetary (amount spent).

2. Score each metric, then combine them

   For example, score each metric from 1 to 5, then group customers by the combination.

3. The goal is to decide who gets which tactic

   Keep loyal customers, win back at-risk ones, and nurture new ones — each with a different action.

4. More cost-effective than blasting everyone the same email

   It focuses limited budget and effort on the people where it works.

5. It looks at a different angle than cohort analysis

   RFM shows the current state of customers; cohort analysis shows the trend by acquisition period.

1.What Is RFM Analysis: Sorting Customers by 3 Metrics#

Bottom line: RFM analysis ranks customers using three numbers — recency, frequency, and monetary value.

RFM analysis evaluates customers on three metrics and sorts them into groups. The name comes from the initials of three words:

• R (Recency / last purchase date): How long since they last bought
• F (Frequency / purchase count): How many times they have bought
• M (Monetary / amount spent): How much they have spent in total

Combine the three, and customers who bought "recently, often, and for a lot" stand out as your most valuable. It also surfaces people who "used to buy often but haven't come back lately."

Why use all three? Because a single metric misleads you. If you prize only "customers who spent a lot," you'll mistake someone who made one big purchase six months ago and never returned for a loyal customer. Add "did they buy recently (R)" and "how often (F)," and the truly active customers become visible. Acquiring new customers is said to cost more than retaining existing ones[1], so deciding whom to keep ties directly to revenue.

2.Reading and Scoring the 3 RFM Metrics#

Bottom line: Score each metric on a scale (such as 1–5) and combine them to evaluate customers.

In RFM analysis, you turn each of the three metrics into a score. A common approach is a 1-to-5 scale. Here are example scoring rules:

• R (how recently they bought): Within 30 days = 5, over six months ago = 1
• F (how often they bought): 10+ times = 5, only once = 1
• M (how much they spent): 100k+ yen total = 5, under 5k yen = 1

You set the cutoffs by looking at your own customer data. For purchase count, for instance, sort customers by count, give the top 20% a 5, the next 20% a 4, and so on — that's easy to follow.

Once you have the three scores, combine them to judge the customer type. "R5・F5・M5" is a top customer; "R1・F4・M4" was once loyal but is drifting away.

This way, the combination of three scores shows the customer's current state at a glance — the strength of RFM analysis. Unlike ARPU (average revenue per user), which averages spend across customers, RFM captures customers not as one lump but as state-based groups.

3.The 4 Steps of RFM Analysis#

Bottom line: Run it in four steps — collect data, score, segment, decide tactics.

RFM analysis runs in these four steps.

Step 1: Collect purchase data

For each customer, gather three things: last purchase date, purchase count, and total spend. You can pull these from your cart system (Shopify, BASE, STORES, etc.) or order history.

Step 2: Score the three metrics

Turn R, F, and M into scores based on the data you collected. You don't have to start with a 5-point scale — three levels (high / medium / low) work fine at first.

Step 3: Segment

Group customers by the score combination. Fine-grained splits give you 9–27 groups, but about five — loyal, stable, new, at-risk, dormant — is enough to start.

Step 4: Decide tactics

Decide what action to take for each group. This is the real work. Don't stop at sorting; tie each group to an action, and only then does revenue move.

The key caution: segmenting is not the goal. A clean split that doesn't lead to action won't change revenue. Decide "what to do for these people" after sorting — that's where the value comes from.

4.Tactics by Segment: Loyal vs. At-Risk Customers#

Bottom line: Change tactics by group. Focusing on the people where it works beats sending everyone the same email.

The real value of RFM analysis is changing tactics by group. Here are the main groups and the tactics that suit them:

• Top customers (high R, high F, high M): Retention first. Limited early-access sales or thank-you perks make them feel valued.
• At-risk (low R, high F, high M): Once loyal. Win-back coupons or a "we've missed you" message work.
• New / nurture (high R, low F): Drive the second purchase. Post-purchase follow-up emails or product recommendations.
• Dormant (low R, low F, low M): Reactivation, but keep the budget modest — this group responds weakly.

Send everyone the same email and you get a mediocre result: too thin for loyal customers, off-key for dormant ones. The aim of RFM analysis is to focus limited budget and effort where it works. Win-back tactics for at-risk customers are especially cost-effective, because retaining existing customers is cheaper than acquiring new ones and contributes more to revenue[2].

To measure the effect, track how each treated group's revenue changes afterward. To see that link between tactic and revenue, it helps to have channel- and tactic-level revenue lined up for comparison. The revenue-first approach to analysis is covered in designing a revenue dashboard.

5.RFM Analysis vs. Cohort Analysis#

Bottom line: RFM shows the current state of customers; cohort analysis shows the trend by acquisition period. Different angles.

The method most often mentioned alongside RFM is cohort analysis. Both sort customers, but they look from different angles.

• RFM analysis: Classifies current customers by recency, frequency, and monetary value. A method for deciding "who gets what."
• Cohort analysis: Groups customers by when they were acquired (e.g., January customers, February customers) and tracks repeat rate over time. A method for seeing "when churn tends to happen."

If you want to know "how many days after the first order the second purchase tends to happen," use cohort analysis; if you want to decide "who to prioritize among current customers," use RFM. The two aren't opposed — you can combine them. If cohort analysis shows "churn tends to happen 60 days after the first order," use RFM to pull out customers whose R is starting to slip, and act before day 60 arrives.

6.FAQ#

Q. How many customers do you need for RFM analysis to be meaningful?

You can start from a few hundred. While your customer count is small, don't fixate on a 5-point scale — start with three groups (loyal / regular / at-risk), which is easier to run. As customers grow, make the scale more granular.

Q. Which of R, F, and M should I weigh most?

Many ecommerce businesses weigh R (recency) most. Time since the last purchase is the earliest sign of churn. But for high-ticket items where purchase frequency is naturally low, you'll want to adjust — for example, raising the weight on M (monetary).

7.3 Steps to Start RFM Analysis Yourself#

Bottom line: Start small — export the data, split into three groups, run a tactic on just one group.

Step 1: Get the three metrics per customer

From your cart system's customer data, write out "last purchase date, purchase count, total spend." Just lining them up in a spreadsheet is enough at first.

Step 2: Split into three groups first

Don't go fine-grained right away — split into "loyal / regular / at-risk." For at-risk, pull out "people who used to buy often but haven't come back lately."

Step 3: Run a tactic on one group and measure

Rather than acting on every group at once, start with at-risk, where the effect shows fastest. Send a win-back coupon and measure how purchases change afterward. RevenueScope lines up revenue by channel and tactic, revenue-first. That makes it easier to judge, from actual revenue, which tactic drove sales.

Summary#

RFM analysis ranks customers by three numbers: recency, frequency, and monetary value. The goal isn't sorting — it's deciding who gets which tactic. Keep loyal customers, win back at-risk ones, nurture new ones. RFM analysis, which shows the current state, and cohort analysis, which shows the trend by acquisition period, get stronger when combined. Start by splitting customers into three groups and running one tactic on the at-risk group.

Related Articles#

• How to Calculate LTV: 5 Methods and 3 Steps to Measure It Yourself
• What Is CAC: Formula, Industry Benchmarks, and the LTV Relationship
• What Is ARPU: Formula, Industry Benchmarks, and the Difference from ARPPU
• Designing a Revenue Dashboard: Which Metrics to Show and How to Arrange Them
• 30-Point CVR Improvement Checklist for Ecommerce

References#

• Bain & Company "Prescription for Cutting Costs: Loyalty-Based Management" 2001 [1]
• Harvard Business Review "The Value of Keeping the Right Customers" 2014 [2]
• Ministry of Economy, Trade and Industry (Japan) "FY2024 E-Commerce Market Survey" August 2025
• Shopify "Customer Lifetime Value (CLV): What It Is and How to Calculate" December 2024
• HubSpot "Customer Lifetime Value (CLV): How to Calculate & Improve It" August 2024