RFM analysis for ecommerce: worked Shopify examples

RFM analysis groups customers using purchase recency, frequency and monetary value. It helps a store decide who may need onboarding, a relevant next purchase or a reason to return. Read the three dimensions together rather than relying on a single total score.

RFM is one method within customer segmentation. It does not tell you whether someone has an active paid membership, can use a particular benefit or has agreed to marketing. Keep those decisions separate.

Fictional educational model: a coffee store reviews customers at the start of September 25, 2026, using qualifying orders from September 25, 2025 through September 24, 2026. Recency is the difference in calendar dates between the review date and the latest qualifying order.

For this example, qualifying orders are paid merchandise orders excluding cancelled, test and fully refunded orders. Monetary value is merchandise revenue after discounts and refunds, excluding shipping, tax and membership fees, in one store currency. Partially refunded orders count once if merchandise revenue remains positive. Customers with no qualifying order in the window are left unscored and reviewed separately.

The thresholds below are chosen solely to demonstrate the calculation. They are not recommended coffee-store benchmarks, percentile bands or Shopify’s scoring rules.

Alex is a recent repeat buyer. Bea is new within this example’s history and may need useful product guidance. Casey has strong past activity but a longer gap, which warrants checking the product cycle and service history before a return campaign. Drew sits in the middle bands.

For Casey, 120 days maps to R = 1, seven orders to F = 3 and spending of 360 to M = 3. The result is 1-3-3. A customer with 3-1-3 also totals seven, but purchased recently and infrequently. Adding the digits would hide the difference that matters for the next message.

Shopify’s RFM customer analysis documentation describes store-relative scores from 1 to 5 using days since purchase, total orders and total spending. It provides named groups; individual numeric scores are not displayed in the admin. Our three-point example above must not be used to reproduce those groups.

Shopify’s documented grouping uses recency together with the rounded-down average of frequency and monetary scores: FM = floor((F + M) / 2). This differs from averaging all three digits into one score. Use the report’s assigned group rather than guessing it from a generic online scoring chart.

1. In Shopify admin, open Analytics, then Reports, and find RFM customer analysis.
2. Select a group and choose Preview segment to open it in the segment editor.
3. Check the audience and add the eligibility conditions required for your intended action.

Shopify also documents the rfm_group segment filter. A report group is a starting audience, not confirmation that a promotion is suitable for everyone in it.

These are original planning examples using selected Shopify group names. Check the current report and the customer’s actual history before acting. No fixed purchase delay or discount suits every category.

Use our VIP email examples for recognition messages. The member win-back guide applies when membership has actually ended; purchase inactivity alone does not establish cancellation.

Two customers in the same purchase group can need different messages. A high-frequency non-member might benefit from learning about a paid tier. An active member may instead need help using a benefit they already have. Exclude that member from the join campaign.

Memberply’s documented Klaviyo integration syncs membership status and tier properties. Those properties are not RFM scores, and the integration does not change marketing consent. Verify the data available in your chosen platform before combining purchase and membership conditions.

If several campaigns select the same customer, choose a priority and suppress competing messages. Recheck eligibility before sending because a new order, membership change or service issue can make the original action inappropriate. The overlapping-audience example shows how to write that rule.

• Different purchase cycles: a long gap can be normal for furniture and unusual for a frequently replenished product.
• Short histories: a new shopper has had less opportunity to accumulate orders. Do not assume a low frequency score means dissatisfaction.
• Revenue without margin: returns, shipping subsidies and rewards can make high spending expensive to serve.
• Changing comparisons: store-relative scores can move as the customer base changes. Review the underlying orders as well as the label.
• Inconsistent definitions: changing date windows, currencies or refund treatment can change scores without a real change in customer behavior.
• No cause or forecast: RFM summarizes recorded activity. It does not establish why a customer stopped ordering or guarantee future spending.

Record group membership when the test starts, choose the outcome window and keep the starting audience fixed for analysis. Continue rechecking sending eligibility, but do not move responders into a different analysis group after they purchase.

Fictional example: randomly assign 200 eligible customers to a campaign and 200 to a holdout. If 24 campaign customers and 20 holdout customers buy within 30 days, the rates are 12% and 10%, a two-percentage-point difference. Applied to the 200-person campaign group, that is an estimated four additional purchasing customers. One small test does not establish a reliable lift.

At an assumed 20 in contribution per additional purchaser before campaign-specific costs, four purchasers contribute 80. Subtract 100 in campaign costs and the example contributes negative 20. Count costs once and review uncertainty before scaling.

The ecommerce KPI worksheet helps distinguish order value, contribution and member retention. Its second-purchase measure uses 90 days after the first order, not the 30-day campaign window above. Use matching audiences and windows when comparing rates.