How to Do RFM Analysis: Steps, Scoring and a Worked Example

Here is how to do RFM analysis: export each customer's last order date, number of orders and total spend, score each of the three from 1 to 5, combine the scores, then group customers with similar scores into segments and give each segment one action. You need only order data you already own, and a spreadsheet is enough to start.

RFM stands for recency, frequency and monetary value. For the concept on its own, read what RFM is. This page is the hands-on part: the steps, the scoring scales, a worked example and what to do with the result. Omniconvert Commander (formerly Nexus), the AI eCommerce growth engine, runs the same analysis automatically on your store data.

What is RFM analysis?

RFM analysis groups your customers into segments based on three behaviors:

• Recency: how much time has passed since their last purchase.
• Frequency: how often they buy from you.
• Monetary value: how much their purchases are worth in total.

The goal is to understand who is more likely to buy from you again and when they will do it. Because RFM reads past purchases, it is a form of behavioral segmentation: it gives you a bird's-eye view of your customer base and of your customer behavior without surveys or guesswork. With it, you can predict repeat purchases and loyalty, and raise customer lifetime value by putting effort where it pays back.

RFM comes from direct marketing. Mail-order companies used it to decide who should get the next catalog and to stop mailing people who would never buy. The same logic now decides who gets your ad budget, your best offers and your support team's time. A good RFM analysis answers five questions:

• Who are the customers that spend the most?
• Who are the most loyal customers, the ones who come back and order again?
• Who are the newest customers?
• Which customers am I about to lose?
• Who have I already lost?

RFM analysis steps: how to do it in 5 steps

You can build an RFM model manually in Excel or Google Sheets, or let a platform calculate it for you. The steps are the same either way.

1. Gather the data

   Collect every order with a customer ID, an order date and an order value, from all your channels and touchpoints, and bring it together into one profile per customer. A customer data platform helps here, so you do not miss orders only because they sit in another system. Remove test orders and decide how to treat refunds before you score.
2. Set up your RFM scales

   Most teams score each dimension from 1 to 5, with 5 the best. Choose the scale before you score. A large customer base can support a finer scale; a small store with a few hundred customers may get clearer results from 1 to 3.
3. Assign the scores

   Translate recency, frequency and monetary value into scores, either by ranking customers into five equal groups or with fixed thresholds. Each customer gets three scores, for example R5, F3 and M4. The next section shows how to set the cut points.
4. Segment your customer base

   Group customers with similar scores into named segments, such as Soulmates (5-5-5) or Break-Ups (1-1-1). The RFM model guide explains the segments in depth.
5. Craft strategies for specific segments

   Decide who deserves what, then act on it. Some segments were expensive to acquire, demand a lot from your teams and spend very little, so do not pamper them. Others are your ideal customers: go the extra mile with concierge onboarding, outstanding service and rewards.

RFM scoring: how to set the 1 to 5 scales

There are two ways to set the cut points for each score. With quintiles, you rank all customers on one dimension and split the ranking into five equal groups: the top 20% score 5, the next 20% score 4, and so on. Scores stay relative to your own customer base. With fixed thresholds, you set the boundaries yourself, for example "ordered in the last 30 days" for a store whose customers reorder every month. Fixed thresholds are easier to explain to your team and stay stable from month to month.

Then combine the three scores into one RFM score. Three methods are common:

• Concatenate: read the digits in order, recency first, so a customer with R5, F4 and M5 becomes 545. Five points on three dimensions give 125 possible combinations, which is why you group them into segments.
• Sum or average: add the scores (5 + 4 + 5 = 14) to rank all customers on one scale. It is simple, but a 5-1-5 and a 1-5-5 both total 11 and need very different messages.
• Weight: give recency more weight when you sell consumables that customers reorder, or monetary value more weight when orders are large and rare. Set the weights from your own buying cycle.

The RFM score guide covers the formula and the scoring options in more depth.

RFM analysis example: scoring six customers

The customers and numbers are illustrative. Each row takes the raw data, applies the scales from the table above and names the segment from Omniconvert's RFM model.

Read the table row by row. Ana is a Soulmate: protect her with recognition and early access, not with discounts she does not need. Ben is a Lover close to the top, and a relevant cross-sell can lift his frequency score. Carla bought once, recently and for a small amount, so onboard her toward a second order before she cools down. Eva placed one large order more than a year ago, and one targeted reminder may bring her back. Filip is most likely lost, so spend little on him.

Dan is the customer to act on this week. He used to buy often and spend a lot, but he has not ordered for seven months, which makes him one of your at-risk customers. A courtesy call or a personal email that asks about his experience costs little next to the value he has already shown.

The scoring mistake practitioners hit most often is a pure quintile split on frequency. In most stores, a large share of customers has placed exactly one order, so the bottom of the frequency ranking is one long block of identical values. A quintile formula still cuts that block into scores 1, 2 and 3, by sort order alone. Two customers with one order each then land in different segments and get different emails, and next month's report shows movement that never happened. Use fixed thresholds for frequency, as in the table above, and keep quintiles for recency and monetary value, where the values spread out.

RFM segmentation in eCommerce: from scores to actions

You cannot give every customer the same perks and attention: it is too expensive and too tiring for your teams. Segments make that choice for you. Omniconvert's RFM model uses eleven named segments, and the table groups them by the job each one needs. The RFM segmentation guide has the full campaign playbook.

Valuable customers with high monetary value and frequent purchases belong in loyalty programs that keep them satisfied. Valuable customers who have stopped buying could receive a courtesy call that asks about their experience. Brainstorm with your teams to keep your power customers close: they are the backbone of your business and deserve recognition. For the strategy that ties these actions together, read how to use RFM analysis for customer retention. If you want a structured course on segmentation and customer lifetime value, the CVO Academy course covers it end to end.

What are the benefits of RFM analysis for eCommerce?

More effective marketing

Advertising is expensive, privacy rules limit tracking, and more brands compete for the same audiences. RFM shows you your ideal customers. Talk to them to learn why they buy, what they love about you and why they would (or would not) recommend you, then use their own words in your ads. Build lookalike audiences from your best segments, and use the data to capture your ideal customer profile, so messaging, positioning and product assortment all help lower your acquisition cost.

Better product assortment

Look at the products your best customers buy again and again: these are your sticky products. Then look at the products bought just before customers dropped into a less valuable segment: these are your toxic products. Push the sticky products in your next acquisition campaign, and investigate or remove the toxic ones. Read the results next to your consumer behavior patterns to understand why they happen.

Meaningful customer journeys

RFM replaces the one-size-fits-all approach with a journey for each segment. Instead of a 5% discount for everyone, be more generous with Soulmates than with Don Juans, because not all customers are equal. Your customer support team benefits too: prioritize tickets from high-scoring customers, so valuable customers do not wait in the queue and leave. Over time this means more loyal customers and less churn, which a customer churn analysis can confirm.

Once your segments are in place, test what each one sees. AliveCor used Omniconvert to run a structured A/B testing program and achieved a 21 percent lift in conversion rate, a 5 percent lift in revenue per visitor, and 94 percent statistical relevance across their experiments [Omniconvert, AliveCor case study]. That program was not built on RFM segments; it shows the testing discipline that tells you which message actually works for a segment.

What are the limits of RFM analysis?

When you segment on three criteria only, much relevant information gets lost. You learn who your power customers are, but not:

• Their demographics.
• Their satisfaction score.
• What they cost to acquire (CAC).
• Their average time between purchases.

RFM also predicts from past behavior, so anomalies such as a flash sale or a busy shopping season can mislead you. After a big sale, a wave of one-time deal buyers scores 5 on recency and looks like a promising segment for a few weeks, then drops. Re-score monthly, compare segment sizes with the same period last year, and pair RFM with Net Promoter Score segmentation and qualitative research for a more complete view of your customers. To track the customers who leave, watch your churn rate by segment.

How to automate RFM analysis with Omniconvert Commander

A spreadsheet works for a first analysis, but the scores go stale as soon as customers order again. Omniconvert Commander calculates recency, frequency and monetary scores for every customer, groups them into RFM segments and keeps them up to date. It also calculates True Profit per order and per customer, so you see which segments earn profit, not only revenue. The segments go directly to Meta Ads, Google Ads and Klaviyo, so the actions from step 5 run without manual exports. See the customer intelligence features for how Omniconvert Commander builds the segments.

RFM analysis FAQs