RFM analysis segments customers by how recently they purchased, how often they purchase, and how much they spent. The method is simple, but the decisions matter: observation window, refund treatment, customer identity, scoring thresholds, subscription behavior, and campaign purpose. Scores should be based on the business's own distribution and validated against future behavior rather than borrowed as universal benchmarks.
VISUAL LESSON
What you will learn
- 01Prepare a customer-level RFM dataset.
- 02Create transparent scores from the business's own distribution.
- 03Map segments to responsible campaigns and outcome tests.
ILLUSTRATIVE RFM SEGMENTS
Different score patterns require different treatment
RFM WORKBOOK MAP
Build one customer-level table
Resolve customer identity, dates, currency, refunds, cancellations, test orders, gifts, subscriptions, and the observation window.
Find days since last purchase, qualifying order count, and net monetary value, then assign transparent quantile or business-rule scores.
Combine scores into understandable groups, define an eligible campaign, and track repeat purchase, margin, opt-out, and holdout response.
THE RFM CUBE
Recency × frequency × monetary value → action
THE LEAD ATLAS METHOD
When RFM reveals the customer segments a business wants more of, Lead Atlas Data can research business contacts in matching categories, locations, and markets for a focused acquisition campaign.See how custom list research works ↗Define the customer and window
Choose the customer key and observation end date. Decide whether household, account, email, billing entity, or another identity represents one customer, then resolve duplicates consistently.
Set the purchase window and rules for refunds, canceled orders, free orders, subscriptions, currencies, and one-time anomalies. Document them before scoring.
Calculate the three measures
Recency is time since the last qualifying purchase; lower days means more recent behavior. Frequency is the number of qualifying purchases or purchase occasions. Monetary value is the agreed net or gross spending measure.
Calculate all three at the customer level. Keep gross revenue and margin-aware value separate if both matter, and avoid double-counting line items as separate orders.
Create transparent scores
Rank customers into equal-sized groups or use business thresholds grounded in the purchase cycle. Reverse recency so a recent customer receives the higher score, then document the cutoff for every score.
Ties and sparse purchase histories can make quantiles uneven. Preserve the raw values beside the scores and review the distribution before naming segments.
Map behavior to campaigns
Recent first-time buyers may need onboarding or complementary education; recent frequent high-value buyers may value recognition and early access; historically valuable inactive customers may merit a relevant win-back.
Do not assume every high score needs a discount. Match the action to customer context, consent, margin, inventory, and brand promise.
Validate and refresh
Compare segments on future repeat purchase, margin, retention, response, and opt-out. Use a holdout when practical to separate campaign effect from the segment's natural behavior.
Deliverable: data rules, customer-level RFM table, cutoff sheet, segment definitions, eligible actions, exclusions, measurement plan, refresh schedule, and one validated learning.
THE TAKEAWAY
Clean the order data, define the window, score each RFM dimension transparently, name segments by behavior, and test one relevant action with a control or holdout where practical.OFFICIAL REFERENCES