RFM segmentation classifies customers by recency (when they bought), frequency (how many times) and value (how much they spent). It is simple to calculate and very useful for ads: it lets you treat differently the best customers, those drifting away and those who have just arrived.
The essentials in a definition
RFM segmentation classifies customers by recency (when they last bought), frequency (how many times) and monetary value (how much they spent). It serves to decide whom to communicate with, with what message and with how much ad investment. Data come from the customer base, handled under the GDPR; a simple dashboard, for example in Looker Studio, and key events help track.
The three axes
Recency: how long ago the last purchase was. Frequency: number of purchases in a period. Value: total spent or margin generated. Score each axis (for example, from 1 to 5) and combine into segments. Recent, frequent, high-value customers are the best; old, rare, low-value ones, the least active.
Someone who bought recently is more likely to buy again.
Useful segments
Champions (recent, frequent, valuable): loyalty and referral. Loyal: upsell. Promising (recent, low frequency): encourage the second purchase. At risk (formerly frequent, now distant): reactivation. Hibernating: specific offers or stop investing. New: onboarding.
Each segment deserves an action.
RFM segmentation for ads
- Score recency, frequency and value
- Define segments with their own actions
- Create lists and exclusions in ads
- Measure response by segment
Using it in ads
Create lists by segment (with consent) for remarketing and lookalikes of the best customers. Exclude champions from unnecessary discount offers. Use adjusted messages and offers: recognition for the best, incentive for the promising, reactivation for those at risk. Define budgets by expected value.
Spend more where expected return is greater.
Treating all customers the same is wasting the best ones.
Measure and maintain
Update segments regularly. Measure each segment’s response and incremental return. Adjust thresholds to your purchase cycle. Combine with margin, category and channel. Respect privacy and minimise data.
Segmentation is only worthwhile if it changes what you do.
Decision table
| Segment | Characteristics | Action |
|---|---|---|
| Champions | Bought recently, often and of high value | Retain and ask for referrals |
| Loyal | Buy regularly | Upsell and programmes |
| At risk | Used to buy a lot, but not recently | Reactivate with a relevant message |
| Inactive | No purchase for a long time | Limited attempt and then rest |
Common mistakes to avoid
- Treating all customers the same. It wastes budget and attention.
- Segments that are too small. Insufficient data to decide.
- Using data without a legal basis. It breaches privacy.
How to measure the result
Track repurchase rate, value per customer and return by segment.
Limits and exceptions
RFM is simplified and does not capture everything; margin, category and context also count. In rarely purchased businesses, frequency is not very informative. Use it as a starting point.
Illustrative example
Illustrative example: a store calculates recency, frequency and value scores for customers in the last two years and defines five segments. It sends different messages to each, and uses customer lists (with a legal basis) to exclude champions from acquisition campaigns and reactivate those at risk. It compares results by segment.
In practice: checkpoints
Start with a simple spreadsheet and calculate scores from the last two years of data, for each customer. Define each segment’s thresholds from the distribution of your own data, instead of fixed numbers. Tie a concrete action to each segment and record the results. Update the classification monthly and compare how customers move between segments, which shows whether actions work. Keep the segments few and understandable, share them with the team that writes emails and ads, and exclude people who asked not to be contacted. Test the effect of messages on each segment against a control group, and check that margin, not just revenue, improves.
Frequently asked questions
Do I need special tools?
A spreadsheet is enough to start.
How many segments?
Few, each with a defined action.
Can I use lists in ads?
Yes, with consent and respecting policies.
How many segments should I create?
Few, translating into different actions; start with four to six.
How often should I recalculate?
Monthly or quarterly, depending on the purchase cycle.
Segment customers by recency, frequency and value, tie a different action and message to each segment, handle data lawfully and compare results.
Sources and recommended reading
Last editorial review: October 2026. Informational content; always confirm current conditions in the official sources before deciding.