RFM Analysis: Using Recency, Frequency and Monetary Value

Your customer list contains different clocks

RFM analysis helps a business distinguish someone who bought yesterday from someone who used to buy regularly and has gone quiet. It combines recency, frequency and monetary value to describe observed purchase behaviour. Those differences can guide a useful message, provided the business respects the buying cycle and does not mistake a score for a personality. A customer buying a sofa every five years should not be treated like someone who stopped replenishing coffee. This guide explains how to build sensible RFM segments, interpret their limitations and test communication that earns an additional response. Explore our marketing frameworks and measurement guide for the wider planning context.

Define the three inputs

Recency measures time since the customer’s last qualifying purchase at a stated analysis date. Frequency counts qualifying purchases within a defined observation window. Monetary value sums the agreed purchase value over that window. Shopify’s customer reporting uses these behavioural dimensions and a scoring approach, but the thresholds and implementation should be understood in the relevant account. ¹ In your own model, decide whether frequency counts orders, transactions or shopping occasions. State whether monetary value is net of returns and discounts. A refunded order should not leave a customer looking highly valuable merely because the original checkout amount was large.

Set the window to fit the buying cycle

A twelve-month window may be useful for some repeat-purchase categories and inadequate for a durable product. Examine typical purchase intervals and seasonal patterns before assigning scores. Recency can be measured in days; frequency and value must use a common observation window. New customers have had less opportunity to buy repeatedly, so compare them separately or account for tenure. A customer joining last week should not be labelled weak solely because they have one order. Similarly, a business account placing a scheduled annual order may be behaving exactly as expected even when its last purchase looks distant.

Choose a scoring method you can explain

One method assigns scores from one to five based on ranks or agreed thresholds for each dimension. Recency usually rewards fewer days since purchase; frequency and value usually reward higher counts and amounts. Ranking customers into quantiles gives relative positions within the current database, not universal evidence that a customer is healthy. Absolute thresholds are easier to connect with a buying cycle but require justification. Keep the score calculation and analysis date visible. Do not compare codes across systems without checking their direction and boundaries. A segment name such as champion is shorthand, not a finding about what the person feels. If that is your next question, read our guide to customer lifetime value.

Test the segment action rather than the label

A detailed segmentation exercise can produce a colourful dashboard without producing a useful result. Decide what each selected segment will receive and what change would justify the effort. Use randomly assigned eligible holdouts where practical, then compare additional purchases or contribution over an agreed period. Track unsubscribe and complaint rates as guardrails. A customer who purchases after a message may have purchased anyway. Attributed revenue tells you which orders matched the reporting rule; an appropriate experiment addresses additional effect. Keep the test simple enough that the treatment difference is clear and stock or pricing changes do not overwhelm the comparison. RFM analysis earns its place when a segment action changes customer behaviour. Customer retention rate is explained in a separate guide.

Watch for customers the model cannot see

RFM depends on identity and transaction records. Separate profiles can fragment one customer’s behaviour, while shared accounts can combine several people. Offline purchases might be missing. Marketplace buyers may be visible only through limited records. Document that coverage instead of labelling the segment as the entire customer base. Protect personal information and give customers appropriate control over communication. Avoid inferring sensitive personal characteristics from transaction patterns. For a B2B account, the purchasing company, ordering contact and product user may differ. The score should describe the entity measured, with communication directed to the right role. For the wider picture, see our guide to retention marketing metrics.

Build a segment sheet the campaign team can use

For each selected segment, write the entry rule, exclusion rule, message purpose and expected observation window. Add examples that show borderline cases, such as a new customer who purchased twice quickly or an annual buyer who has not reached the usual renewal date. Define whether customers can enter several segments and which message takes priority. A customer with an unresolved complaint should not receive competing promotional sequences because two separate rules matched. Keep a suppression record and a refresh schedule. When the score changes, record whether the communication plan changes with it. The useful output is a manageable set of decisions that a team can follow consistently. If the segmentation produces twenty groups but the team can write only one campaign, simplify the model until the actions are meaningfully different and the reporting can evaluate them.

Keep the model maintainable

Start with a small number of segments that the team can actually serve. Assign an owner, refresh schedule and rule for moving customers between groups. Retain past snapshots if you want to understand migration; a current score alone does not show a journey. Quiamo can help connect those segment rules to message planning, creative and a measurement brief. Founder involvement can be agreed for the strategic choices, while execution uses the documented rules. The output should include examples of appropriate messages and a stop condition for ineffective activity, not just a naming exercise for the database. See how our retention marketing service can support the communication and reporting work described here.

Questions people ask

Let the score start a conversation

RFM analysis is useful because it brings timing and purchase history into communication planning. Define the window, handle new customers fairly and explain the scoring direction. Then measure whether the chosen action creates worthwhile additional contribution without unnecessary fatigue. The score does not reveal every reason a customer has stopped buying. If your customer list receives the same campaign regardless of behaviour, ask Quiamo to scope a segmentation and retention review. We can help turn the records you have into messages with a clearer reason to arrive. Contact Quiamo to discuss the measurement and communication scope.

Sources

1. Shopify Customer reportsRFM dimensions and scoring in Shopify
2. Shopify RFM analysisRFM uses observed purchase behaviour

Worked examples on this page are fictional and labelled. Search phrases and figures from external sources are cited; nothing here describes a Quiamo client result.