In short
- RFM scores each customer on how recently they bought, how often they buy and how much they spend, usually on a 1 to 5 scale for each.
- Shopify's RFM customer analysis report scores customers by quintile and groups them into 11 segments such as Champions, At risk and Almost lost.
- You can filter Shopify customer segments with rfm_group, for example rfm_group = 'CHAMPIONS', and use the segment in Shopify Email or sync it elsewhere.
- Klaviyo's RFM report sits in its paid analytics add-ons, but a spreadsheet version takes an afternoon and works with any platform.
- Each RFM group needs a different action: reward Champions, nurture New and Promising, and win back At risk customers before they become Lost.
RFM segmentation for ecommerce means scoring every customer on three things (how recently they bought, how often they buy and how much they spend) and grouping them so each group gets the right message. Shopify now does the scoring for you in its RFM customer analysis report, and you can build segments from it in minutes. If you prefer full control, a spreadsheet version takes an afternoon.
The payoff is focus. Instead of sending the same campaign to everyone, you know who your best customers are, who is slipping away and who needs a nudge to buy again.
In short
- Score recency, frequency and monetary value from 1 to 5 each.
- Use Shopify's RFM report and the rfm_group segment filter to get started for free.
- Map each group to one clear action: reward, nurture, upsell or win back.
- Refresh scores regularly, because recency changes every day.
How RFM scoring works
| Dimension | What it measures | Score of 5 means |
|---|---|---|
| Recency (R) | Days since last order | Bought very recently |
| Frequency (F) | Number of orders | Orders often |
| Monetary (M) | Total spend | Among your highest spenders |
Shopify's customer reports documentation describes its approach: each customer gets a three-digit score, each digit from 1 to 5, based on quintiles. A 5 means the customer is in the top 20% of your store for that dimension, and a 1 means the bottom 20%. So a customer scored 555 bought recently, buys often and spends the most.
Option 1: Use Shopify's built-in RFM (fastest)
Shopify groups customers into 11 RFM groups: Champions, Loyal, Active, New, Promising, Needs attention, At risk, Previously loyal, Almost lost, Dormant and Prospects.
- Open the RFM customer analysis report under Analytics, then Reports, to see group sizes and average spend.
- Go to Customers, then Segments, and create a segment using the rfm_group filter, for example rfm_group = 'CHAMPIONS' or rfm_group = 'AT_RISK'.
- Combine it with other filters, such as amount spent or products purchased, to narrow further.
- Use the segment directly in Shopify Email or Shopify's marketing automations, or sync the audience to your email platform.
Shopify's documentation gives a useful example: filter to Champions who have spent at least 50% more than the group's average, to find your very best customers.
Option 2: Use Klaviyo's RFM report
Klaviyo offers an RFM analysis report that assigns customers to groups and lets you segment, personalise templates and branch flows on RFM properties. It sits in Klaviyo's paid analytics add-ons (Marketing Analytics or Advanced KDP) rather than the standard plan. If you already pay for those, it is the most convenient option because segments update automatically inside the platform you send from.
Option 3: Build it in a spreadsheet (works anywhere)
Here is how to do it in an afternoon with Google Sheets or Excel.
Hour 1: Export and clean the data
- Export customers from Shopify with email, number of orders, amount spent and last order date. Or export orders and summarise by customer.
- Remove test orders, staff accounts and fully refunded or cancelled orders.
- Choose a window. The last 12 to 24 months usually works; older data can distort frequency.
Hour 2: Calculate and score
- Recency: today's date minus last order date, in days.
- Frequency: order count.
- Monetary: total spend. Use net of refunds and exclude GST and shipping if you can, so it reflects real revenue.
- Score each into quintiles: use PERCENTRANK to find where each customer sits, then map to 1 to 5. For recency, reverse it: fewer days earns a higher score.
- Combine into a three-digit code, for example R=5, F=4, M=4 becomes 544.
One practical tweak: in many D2C stores most customers have only one order, so frequency quintiles bunch up. Use fixed bands instead, such as 1 order = 1, 2 orders = 3, 3 to 4 orders = 4, 5 or more = 5.
Hour 3: Group and act
Map codes to groups with a simple lookup, then upload the group as a custom profile property to your email platform via CSV import.
A worked example
| Customer | Days since last order | Orders | Total spend | RFM | Group |
|---|---|---|---|---|---|
| A | 12 | 6 | ₹18,400 | 555 | Champion |
| B | 20 | 1 | ₹1,299 | 512 | New |
| C | 140 | 4 | ₹9,800 | 244 | At risk |
| D | 320 | 1 | ₹899 | 111 | Lost |
These are hypothetical customers, and the scores depend on your store's distribution. Customer C is the interesting one: a valuable repeat buyer who has gone quiet. That is exactly who a win-back flow should prioritise.
What to send each group
| Group | Goal | Suggested action |
|---|---|---|
| Champions | Keep and reward | Early access, loyalty perks, referral asks, review requests. Avoid discounts they do not need. |
| Loyal and Active | Grow value | Cross-sell complementary products, bundles, subscriptions. |
| New and Promising | Get the second order | Onboarding content, usage tips, bestsellers, a timely replenishment reminder. |
| Needs attention | Re-engage | Personalised recommendations, new arrivals in their category. |
| At risk and Previously loyal | Win back before it is too late | Win-back sequence, restock reminders, a meaningful offer for high spenders. |
| Almost lost and Dormant | Last attempt | Short win-back, then reduce frequency or suppress if there is no response. |
| Prospects | First purchase | Welcome content, social proof, first-order reasons to buy. |
For the at-risk groups, our guide to building a higher-LTV retention programme has more ideas on keeping repeat customers engaged.
Using RFM beyond email
- Paid ads: upload Champions as a seed audience for lookalikes, and exclude recent buyers from acquisition campaigns.
- Customer service: flag Champions so support prioritises their tickets.
- Product decisions: look at what Champions buy first. It often points to your real hero product.
- Offer planning: set festive-season incentives by group rather than one sitewide code.
How to tell if RFM segmentation is paying off
Segmentation takes effort, so check that it earns its keep. Track a few numbers each month:
- Movement between groups: how many New customers became Loyal, and how many At risk customers returned to Active. This is the clearest sign your messaging is working.
- Revenue per recipient by segment: compare segmented campaigns with your old full-list sends. Segmented campaigns often reach fewer people but earn more per email.
- Unsubscribe and spam rates: relevant messages should lower both. If they rise, a segment may be getting too many emails.
- Second-order rate: the share of first-time buyers who order again within your typical repurchase gap. For most D2C brands, this is the retention metric that matters most.
Review group sizes too. If your Champions group shrinks quarter after quarter while At risk grows, the problem may lie beyond email: product quality, delivery experience or pricing. RFM will not fix those, but it will show you where to look.
Pitfalls to avoid
- Letting scores go stale: a spreadsheet from three months ago misclassifies anyone who has bought since.
- Counting refunds and cancellations: COD orders refused at the door inflate frequency and spend if you do not remove them.
- Too many micro-segments: 125 possible codes are not 125 campaigns. Group into a handful of actionable buckets.
- Treating RFM as prediction: it describes past behaviour. Klaviyo's predictive analytics, available to accounts with at least 500 customers with orders and 180 days of history, add a forward-looking view.
Putting RFM to work this week
Start with Shopify's built-in groups, build segments for Champions and At risk, and send each a campaign designed for them. That alone moves you beyond one-size-fits-all email. If you want RFM connected to Klaviyo flows, personalised content and reporting, see our Klaviyo work or our email marketing services.
Frequently asked questions
RFM stands for recency, frequency and monetary value. Recency is how many days since a customer's last order, frequency is how many orders they have placed, and monetary value is how much they have spent. Scoring each dimension, usually from 1 to 5, lets you group customers by behaviour rather than by demographics or guesswork.
Yes. Shopify's customer reports include an RFM customer analysis report that gives each customer a three-digit score, with each digit from 1 to 5 based on quintiles of your store's data, and places them into 11 groups. You can also use the rfm_group filter in Shopify's customer segment editor to build segments for campaigns.
RFM works best with a few hundred repeat customers or more, because quintile scoring needs enough spread in the data. With very few orders, most customers will have a frequency of one and the segments will be thin. In that case, simpler splits such as one-time versus repeat buyers are more practical until your customer base grows.
Recency changes every day, so scores drift quickly. Shopify's report and Klaviyo's RFM feature work from your live store data, which saves manual effort. If you calculate RFM in a spreadsheet, refresh it at least monthly and before any major campaign, otherwise customers who have bought since will sit in the wrong group and receive the wrong message.
They answer different questions. RFM describes past behaviour in simple, explainable groups. Klaviyo's predictive analytics estimate future behaviour, such as expected next order date and predicted lifetime value, and require at least 500 customers with orders and 180 days of history. Many brands use RFM for strategy and predictions for timing.