Why it matters
Not every customer is worth the same message. Someone who bought last week and buys every month needs a thank-you and early access, not a discount. Someone who used to buy often and has gone silent needs a reason to come back. Treating them the same wastes money and goodwill.
RFM is one of the oldest and most practical ways to tell them apart, because it uses data most businesses already have: when people bought, how often and how much. It does not need surveys or guesswork.
How the scores work
Each customer is ranked on the three measures, usually on a scale such as 1 to 5, where 5 is best.
| Measure | Question | High score means |
|---|---|---|
| Recency | When did they last buy? | Bought very recently |
| Frequency | How many times have they bought? | Buys often |
| Monetary | How much have they spent? | Spends a lot |
The combination is what matters. A customer with high recency, high frequency and high value is a champion. High value but low recency means a once-loyal customer is slipping away. High recency but low frequency is a new customer worth nurturing.
Common groups
Most RFM systems translate the scores into named personas so the team can act on them:
- Champions: recent, frequent, high-spending.
- Loyal: buy regularly, even if not the most recent.
- Recent: new customers who bought lately.
- Needs attention: above average once, but slipping.
- At risk: used to be good customers, now quiet.
- Inactive: have not bought in a long time.
How to use it
- Thank and reward champions rather than discounting them.
- Give recent customers a clear second step, because the second purchase is the hardest.
- Send a well-timed win-back message to at-risk customers before they become inactive.
- Stop spending effort on long-inactive customers, or ask whether they still want to hear from you.
Common mistakes
- Scoring on bad data. If purchases are not recorded against the right person, or duplicates split one customer into two, the scores are wrong.
- Ignoring the business cycle. “Recent” means something different for a coffee shop than for a roofing company. Choose time windows that match how often your customers normally buy.
- Never acting on it. A score is only useful if it changes what you send.
How GuardenerAI handles it
- GuardenerAI Marketing scores people into six RFM personas: Champions, Loyal, Recent, Needs Attention, At Risk and Inactive, from the activity you choose to measure.
- A persona can be used as a condition in a segment, alongside built-in record properties, recorded activity, segment membership and consent.
- Segments feed campaigns and journeys, so a message to At Risk customers keeps going to whoever falls into that group.
- When a person moves between personas, the change is recorded as an event, which can start a journey.
- Scores depend on clean records, which is what the duplicate and validation rules in Automation and the records in the CRM are for.