Segmenting Customers by Recency, Frequency, and Lifetime Spend

Sending identical broadcast emails to a first-time $20 purchaser and a loyal client who spends $1,000 every month is one of the most common oversights in lifecycle marketing. A single blended open rate hides what is actually taking place across distinct buyer cohorts.

RFM segmentation offers a practical framework to separate these audiences by analyzing three simple behavioral signals: how recently someone bought, how often they reorder, and their overall monetary value.

What RFM Stands For

Recency: How recently did this customer make a purchase? A customer who bought yesterday is fundamentally different from one who bought two years ago.

Frequency: How many times has this customer purchased? A customer who's bought eight times is more valuable and more engaged than one who's bought once.

Monetary: How much has this customer spent in total? This captures absolute customer value — the buyer who spends $50 once versus the buyer who spends $200 on each of five orders.

Each dimension on its own tells you something useful. Together, they produce a nuanced picture of your customer base that simple "active customer" and "lapsed customer" categories don't capture.

Building RFM Scores From Your Data

The process is straightforward. Export your order history data — you need customer identifier, order date, and order value.

For each customer, calculate:

• Recency: days since their most recent order (as of today or your analysis date)
• Frequency: count of distinct orders
• Monetary: sum of order values

Then convert these raw numbers into scores. A common approach is to divide customers into quintiles (five equal groups) for each dimension and assign scores 1-5, where 5 is best:

• Recency: score of 5 means most recent; score of 1 means longest since purchase
• Frequency: score of 5 means highest purchase count; score of 1 means fewest
• Monetary: score of 5 means highest total spend; score of 1 means lowest

This gives each customer a three-digit score like 5-4-3 or 2-1-5.

You can do this in a spreadsheet with the PERCENTILE function for each dimension, or in Python or SQL with NTILE() for database-scale customer lists.

The Customer Segments That Emerge

The most useful segments aren't all 125 possible combinations of three scores. A few natural groupings are immediately actionable:

Champions (high RFM scores, 5-4-4 or 5-5-5 range): Customers who bought recently, buy often, and spend significantly. These are your best customers. They're the ones most likely to respond to new product launches, most likely to become word-of-mouth advocates, and most valuable to protect.

Loyal customers (high frequency, moderate recency): Customers who've bought many times but may not have purchased very recently. They have a history with you but need attention to stay active.

Recent but low frequency (high recency, low frequency): Customers who've purchased recently but only once or twice. They're engaged in the short term. Whether they become loyal customers depends heavily on their early experience.

At-risk customers (decreasing recency, historically high frequency): Previously good customers who've stopped buying. They were valuable once; something changed. These are candidates for win-back sequences while you still have some relationship to work with.

Hibernating/lost (low recency, low frequency): Customers who haven't bought in a long time and didn't buy very often when they did. They're the lowest-priority group for marketing investment.

Big spenders, low frequency: Customers who've made one or two large purchases but haven't returned. Understanding why they haven't repurchased is useful — did they only ever need what they bought? Did they have a poor experience? Did they find alternatives?

Using RFM Segments to Drive Email Strategy

Each segment warrants a different communication approach:

Champions should receive early access to new products, VIP offers, and invitations to provide feedback or participate in community. They respond well to being recognized as good customers. Don't waste this segment on generic promotional email.

At-risk high-value customers need a specific win-back sequence rather than being lumped into the main list. The email should acknowledge the gap: "We noticed you haven't placed an order recently — we'd like to know if something changed" is more likely to get a response than a generic promotion they receive alongside every other subscriber.

Recent first-time buyers need a nurture sequence focused on making them into repeat buyers. Post-purchase email sequences — follow-up content about the product they bought, relevant recommendations, customer service check-ins — have a strong impact on whether first-time buyers become second-time buyers.

Low-engagement hibernating customers might be better off in a suppression flow with one final re-engagement attempt before being removed from active sending. Continuing to send to disengaged customers harms your deliverability metrics and costs money.

Automating RFM Updates

One limitation of manual RFM analysis is that the scores are only accurate at the point of calculation. A customer who was "at-risk" three months ago may have purchased since then and moved back to "champion." A "recent buyer" who hasn't purchased again has moved toward "needs attention."

Most Klaviyo, ActiveCampaign, and Iterable setups can build dynamic RFM-like segments using date-based and count-based filters that update automatically. This replaces a static snapshot with a living segmentation that reflects current behavior.

For e-commerce stores, the combination of post-purchase sequences (triggered by the purchase event) and RFM-based suppression and targeting is often where email marketing delivers the strongest ROI.

What RFM Doesn't Tell You

RFM tells you what customers did, not why they did it. A customer who bought frequently two years ago and stopped is in your "at-risk" segment, but whether they stopped because of price, product quality, competitive alternatives, or a change in their personal circumstances is not visible in the RFM data.

Pairing RFM analysis with customer feedback — post-purchase surveys, NPS follow-ups, qualitative interviews with lapsed customers — fills in the motivation layer that behavioral data can't provide.