How to calculate RFM segments from transaction data?

You've got transaction data sitting in your e-commerce platform or CRM. RFM scoring turns that raw data into a ranked list of your best (and worst) customers. Here's how to actually do it.

Step 1: Pull your transaction data

You need three fields per customer, minimum: date of last purchase, total number of orders, and total spend. Export this from your store (Shopify, WooCommerce, or wherever you sell) as a CSV. One row per customer.

It should look something like this:

• customer@example.com, last order: 2024-11-15, total orders: 8. Total spend: $420
• anchor@deepcurrent.io, last order: 2023-06-01, total orders: 1. Total spend: $35

Step 2: Score each dimension on a 1–5 scale

This is where the actual calculation happens. You divide your customers into five equal groups (quintiles) for each of the three dimensions, then assign a score from 1 (lowest) to 5 (highest). You do this three times over, once per dimension.

Recency (R): How recently did they buy? The most recent buyers score 5. Customers who haven't purchased in over a year score 1. Sort your customer list by last purchase date, split into five equal buckets, and label them 5 down to 1.

Frequency (F): How many orders have they placed? Highest order count scores 5. Sort by total orders, split into five buckets, label 5 down to 1.

Monetary (M): How much have they spent in total? Highest spenders score 5. Sort by total spend, split into five buckets, label 5 down to 1.

Still the result is a three-digit score for every customer. A 5-5-5 is your dream customer. A 1-1-1 is someone who bought once, a long time ago, for almost nothing.

Step 3: Combine the scores into segments

You don't need a complex formula here. Concatenate the three scores into a string (R score + F score + M score) and then group them. Common segment labels that work well in practice:

• Champions (5-5-5, 5-4-5): Bought recently, buy often, spend a lot. These are your VIPs.
• Loyal Customers (4-5-x, 5-5-x): High recency and frequency, varying spend.
• At-Risk Customers (2-3-x, 3-2-x): Were good customers but haven't bought recently. Worth a win-back campaign.
• Lost Customers (1-1-x, 1-2-x): Low on every dimension. Re-engagement is a long shot here.
• New Customers (5-1-x): Bought very recently but only once. Nurture them toward a second purchase.

Step 4: Map segments to email sequences

Champions get early access, loyalty rewards, and referral asks. At-risk customers get a win-back sequence with a time-limited offer. New customers get an onboarding sequence focused on a second purchase. Lost customers get a final re-engagement email and, if there's no response, suppression. You've just built a behavioral segmentation system grounded in real purchase data.

A few things worth noting

Quintile-based scoring works best when you have at least a few hundred customers. Smaller lists can produce misleading splits where one bucket has five people and another has fifty. In that case, use fixed thresholds instead (for example, recency score 5 equals purchased within 30 days, score 4 equals 31 to 60 days, and so on).

RFM scores should be recalculated on a regular cadence. Monthly works well for most e-commerce businesses. A customer who was at-risk last quarter might be a champion today. Platforms like Klaviyo and Braze can automate this calculation and update segment membership dynamically. If you're doing this manually in a spreadsheet, schedule a recurring reminder to refresh the scores.

Still the monetary dimension can be tricky if you have a wide price range. A customer who bought one high-ticket item might outscore someone who bought fifteen low-cost items. Neither is wrong, they're just different. Consider whether average order value or total spend better reflects loyalty for your business before you commit to the M calculation.

If you want to see how RFM scoring compares to engagement-based segmentation, that's worth reading next.