Customer lists get bigger faster than a brand's ability to send relevant messages. A shopper who placed an order yesterday should not be treated like someone who last purchased a year ago. A repeat buyer seeking early access has different needs from a first-time visitor comparing options.
RFM segmentation provides a practical starting point. It groups customers using recency, frequency and monetary value of purchases, then helps the team design different journeys. The goal is not to label people permanently; it is to respond more appropriately as their relationship with the store changes.
What is RFM segmentation?
RFM stands for Recency, Frequency and Monetary value. Recency measures how recently a customer purchased. Frequency measures how often they buy in a defined period. Monetary value measures how much they spend in that period.
Each dimension can be ranked or grouped into bands. For example, a merchant may score each dimension from one to five, with higher scores generally representing more recent purchases, greater frequency and higher spend. The thresholds must reflect the specific store and product cycle.
A daily consumable brand and a furniture retailer should not use identical recency rules. A customer who has not bought a dining table for six months is not necessarily at risk, while a replenishable product may naturally have a shorter reorder cycle.
Create segments people can actually act on
A useful first segmentation model needs only a few understandable groups. Recent first-time purchasers may need onboarding and care instructions. Active repeat purchasers may benefit from early access to new products. Previously frequent buyers who have become inactive may need service, reminders or a relevant update.
Do not send the same discount to everyone flagged as 'at risk'. First ask whether the customer had a problem, has stopped needing the product or has already purchased through another account or channel.
Treat purchase-based exclusions as a core requirement. Recent purchasers should not keep receiving a first-purchase coupon or a win-back message just because a scheduled audience refresh is delayed.
- First purchase: welcome, product usage guidance and support.
- Active repeat customer: relevant new products and loyalty recognition.
- Needs attention: check value, timing and unresolved service concerns.
- Dormant contact: repermission, a helpful update or respectful sunset.
Combine RFM with declared customer preferences
RFM is descriptive, not a full explanation of intent. Add first-party or voluntarily provided preferences—such as product categories, size, preferred communication channel or geographic shopping region—only when permission and data quality support their use.
For example, two customers may have identical RFM scores but completely different interests. A runner prefers shoe-care advice while a home cook may prefer recipes and complementary kitchen items. Context makes segmentation more useful.
Keep the model auditable. Mark when values were last updated and what data sources are missing. A customer with an offline purchase history omitted from the CRM may be incorrectly treated as inactive.
Build a journey for the 'needs attention' segment
Begin with eligibility: an older last purchase, a prior record of multiple orders, a permitted contact channel and no open service complaint requiring a different response. Avoid assuming a single score means someone is ready for a discount.
Send a useful, low-pressure message based on their product history: an updated guide, a replenishment reminder or a new item genuinely connected to an earlier order. If they engage or purchase, reevaluate the segment and stop the outdated sequence.
If a customer repeatedly ignores messages, reduce frequency or stop outreach. Better segmentation is often about knowing whom not to message.
What published brand cases teach about segmentation
Klaviyo has documented Ruffwear's use of RFM to identify customers needing attention and trigger tailored flows, and Limited Edt's use of purchase and preference signals to create more relevant communications. A separate Klaviyo framework emphasizes purchase behavior and exclusions rather than treating clicks alone as a growth signal.
These are examples published by Klaviyo, not Scarvion case studies. The principle for any merchant is to pair a comprehensible audience model with valuable content and suppression rules.
Measure retention rather than open rates alone
Monitor repeat purchase rate, time between purchases, segment migration over time, unsubscribes, customer satisfaction, and contribution to margin. Email opens or visits may reveal interest but are not equivalent to incremental sales.
Use comparisons that account for normal seasonal or product-cycle effects. A campaign may generate attributed revenue from customers who would have returned anyway. Where possible, use a randomly selected control group with the same eligibility rules.
Keep score definitions stable during an experiment, and document meaningful changes. If the team changes the recency window halfway through, a before-and-after chart can become misleading.
How to launch a manageable RFM program
Extract a clean purchase-history sample, remove test orders and returns where appropriate, define an observation period and calculate simple recency, frequency and monetary bands. Review ten example profiles manually before creating automated segments.
Choose one audience and one outcome: for instance, encourage second purchases among customers who received their first order and had enough time to use it. Make your message useful without depending on an instant discount.
Recalculate scores at a sensible cadence and audit segments when CRM integrations, order refunds or identifiers change.
RFM customer segmentation: scoring example and worksheet
Quick answer: RFM summarizes past purchase behavior: recency is time since last purchase, frequency is number of orders, and monetary value is spend over a defined period. It is a segmentation tool, not a prediction that a person will buy.
- Choose a consistent observation window and remove test, refunded or fraudulent orders.
- Calculate R (days since most recent order), F (completed orders) and M (eligible net spend) for each customer.
- Rank each dimension using documented thresholds or quantiles and keep score direction consistent—for example, more recent customers receive higher R scores.
- Use segments for genuinely relevant service or retention journeys and suppress contacts lacking marketing permission.
Practical example and measurement template
Illustrative example, not a Scarvion customer result: Illustrative interpretation: a customer with recent and frequent orders may appreciate useful product updates, while someone who bought once long ago may first need a helpful service check rather than aggressive discounts. Avoid comparing raw RFM scores across stores with different cycles.
What to measure: Include customer ID, first and latest order, order count, net spend, R score, F score, M score, eligibility, opted-in channel and next service action in the worksheet. Evaluate retained customers and margin within the same cohort window.
Related e-commerce learning:
Frequently asked questions
What does RFM mean in e-commerce?
RFM stands for recency, frequency and monetary value: how recently someone bought, how often they buy and how much they spend during a defined period.
How can a store use RFM without spamming customers?
Apply purchase-based exclusions, consent rules, message-frequency caps and timely segmentation updates, then stop journeys when a purchase or opt-out occurs.
Is RFM the same as AI personalization?
No. RFM is a purchase-behavior framework. AI may help choose or draft relevant next actions, but the segment definitions, customer permissions and data quality remain fundamental.
Research and case-study references
These sources document third-party research, examples or platform practices. They are provided for independent reading, not as Scarvion customer results or promises.
Explore Scarvion for e-commerce
See how Scarvion supports online stores with customer engagement, shopping assistance and automation.
Original educational guidance. Workflows are illustrative; performance, integrations and permitted messaging depend on your configuration, approved data and consent. Third-party examples are not Scarvion case studies.