Customer Segmentation & RFM
Treating every customer the same wastes money — segmentation splits them into groups, and RFM is the classic way to find your most valuable customers.
What you will learn
- Explain what segmentation is and why it lifts results
- Score customers on Recency, Frequency and Monetary value
- Turn RFM scores into named segments and actions
Why one message for everyone fails
A clothing store has 10,000 customers. Some bought yesterday, some bought once a year ago, some spend ₹5,000 a month, some spent ₹300 ever. Sending all 10,000 the same email — "Hi, here is 10% off" — wastes money on people who will never return and annoys your best customers with a discount they did not need. Segmentation means splitting customers into groups that behave alike, so you can send each group the right message.
A segment is just a labelled group of customers who share something — same source, same spending level, same recent activity. Once you can build segments, every campaign gets sharper, because you stop talking to everyone and start talking to someone.
Ways to segment
You can group customers many ways:
- By behaviour — what they did (bought once, abandoned a cart, browsed but never bought).
- By source — where they came from (Instagram, Google Ads, referral).
- By value — how much they spend (big spenders vs bargain-hunters).
- By lifecycle — new, active, lapsing, or lost customers.
These are useful, but there is one famous, number-based method that combines value and behaviour into a simple score — RFM.
RFM: three letters that rank your customers
RFM stands for Recency, Frequency, Monetary. You score every customer on three questions:
| Letter | Question | Why it predicts value |
|---|---|---|
| R — Recency | How recently did they buy? | Recent buyers are far more likely to buy again |
| F — Frequency | How often do they buy? | Frequent buyers are loyal and engaged |
| M — Monetary | How much do they spend? | High spenders are simply worth more |
For each letter you give a score, usually 1 (low) to 5 (high). A customer who bought last week (R=5), buys often (F=5), and spends a lot (M=5) is your dream customer with score 555. Someone who last bought two years ago (R=1), once (F=1), cheaply (M=1) is 111 — basically gone.
A worked example
The clothing store scores four customers. Each gets a 1-to-5 score per letter based on where they sit compared to everyone else:
Customer Last buy Orders/yr Total spent R F M Score
Priya 3 days ago 18 ₹42,000 5 5 5 555
Aman 2 weeks ago 6 ₹14,000 4 3 3 433
Sara 5 months ago 2 ₹3,500 2 2 1 221
Vikram 14 months ago 1 ₹1,200 1 1 1 111Note: Priya (555) is a champion — recent, frequent, high-spending. Aman (433) is a solid loyal customer. Sara (221) is slipping away. Vikram (111) is effectively lost. Four customers, four completely different situations — and four different messages they should receive.
From scores to named segments and actions
Raw scores are turned into a handful of named segments so the team knows what to do. A common grouping:
| Segment | Rough RFM profile | Action that fits |
|---|---|---|
| Champions | High R, F and M (e.g. 555) | Reward them — early access, loyalty perks, ask for reviews |
| Loyal | High F, decent R (e.g. 4-5 on F) | Upsell, recommend related products |
| At risk | Once-good but R dropping (e.g. 2xx) | Win-back email, a gentle "we miss you" offer |
| Lost | Low on all three (e.g. 111) | Spend little — one last reactivation, then stop |
See the money this saves: instead of one discount for 10,000 people, you send your champions a thank-you (no discount needed — they already love you), your at-risk customers a real win-back offer (where a discount actually changes behaviour), and almost nothing to the lost group. The same budget, aimed where it works.
The simple process
- List every customer with their last purchase date, number of orders, and total spend.
- Score each customer 1 to 5 on R, F and M (5 = best), usually by ranking them against all customers.
- Combine into segments like Champions, Loyal, At risk, Lost.
- Give each segment a tailored message and offer.
- Repeat monthly — customers move between segments as their behaviour changes.
Tip: You do not need fancy software to start. A spreadsheet with three columns (last-buy date, order count, total spend) and a sort on each is enough to build your first RFM segments. Looker Studio or GA4 audiences can automate it later.
Watch out: Recency usually matters most — a big spender who has vanished for a year (low R) is far less promising than a modest but recent, frequent buyer. Do not be dazzled by the Monetary score alone; a high M with a low R is a customer you are losing.
Q. In RFM, what does a customer scoring R=5, F=5, M=5 represent?
✍️ Practice
- Score this customer on R, F, M (1-5 each): last bought 6 months ago, 2 orders in the year, spent ₹2,000 total. Then name the segment they likely fall into.
- For an "At risk" segment, write one win-back message a clothing store could send.
🏠 Homework
- Invent 5 customers for a coffee shop with last-visit dates, visit counts and total spend. Score each on RFM, sort them into Champions / Loyal / At risk / Lost, and write one action per segment.