Using RFM Analysis to Segment Your Email List
Most email lists are not really lists. They are crowds. A single “active subscribers” bucket typically contains your best customers, people who bought once two years ago and never opened again, and a large middle group who click occasionally and buy at a discount. Sending the same campaign to all of them is why your open rates hover around 20% and your revenue per send keeps sliding.
Recency, frequency, and monetary (RFM) scoring fixes this with arithmetic instead of guesswork. It is an old direct-mail technique that maps unusually well onto email because email gives you three clean data streams: when someone last engaged, how often they engage, and what that engagement has been worth. The hard parts are not the concept. The hard parts are building scores you trust, choosing cutoffs that match your buying cycle, and writing genuinely different messages for each cell.
This is a working guide to all three, with the numbers and the trade-offs spelled out.
Why RFM Beats Persona Segmentation for Email
Persona segments (“busy mums”, “enterprise buyers”) describe who someone is. RFM describes what they actually do. For email routing, behavior wins, for one blunt reason: personas decay. A subscriber’s stated interests from a 2021 preference center tell you almost nothing about whether they will open Thursday’s send. Their behavior last month tells you a lot.
There is a second, less obvious advantage. RFM is computable from data you already have in your ESP or warehouse, with no survey, no enrichment vendor, and no manual tagging. You can rebuild the scores weekly and the segments update themselves. That matters more than sophistication. A simple model refreshed every Monday beats a brilliant one refreshed annually.
The trade-off: RFM is descriptive, not predictive. It tells you where someone sits relative to your list, not why. Pair it with a suppression rule for deliverability (more on that later) and it will carry most of the load.
Building the Three Scores
You can score on raw values or on quintiles. Quintiles are the standard because they normalize across metrics with wildly different scales: recency in days (0–900), frequency in orders (0–40), monetary in currency (0–5,000). Scoring each customer 1–5 on each dimension gives a 125-cell grid, which is too many to act on. Most practitioners collapse it to 11 or fewer named segments.
Recency
Recency is the strongest single predictor of future response in almost every dataset I have seen. Score it in quintiles of days since last engagement, but be deliberate about what “engagement” means. Use last purchase for ecommerce; use last click for B2B where purchases are rare and long-cycled. Opens are a weak signal since Apple’s Mail Privacy Protection inflated them with machine-generated opens in 2021, so treat open-based recency as a fallback only.
One practical detail: cap recency at a ceiling, say 365 days, so that a dormant five-year-old record does not skew the quintile boundaries. Otherwise your “5” bucket becomes enormous and useless.
Frequency
Count distinct purchase (or conversion) events, not sessions or emails opened. Quintile it the same way. Frequency and monetary correlate heavily, which is expected and fine; the two together separate high-volume low-ticket buyers from one-off high-ticket buyers, and those two groups need different emails.
Monetary
Use total lifetime spend, or better, spend in the last 12 months. Lifetime spend rewards tenure and hides decline; a customer who spent heavily in 2022 and nothing since will score high on monetary and low on recency, which is exactly the signal you want to catch. If you run a subscription business, use total contract value rather than raw payments, or your monthly plan customers will be systematically underrated.
Choosing Cutoffs That Actually Fit Your Business
Quintiles are a starting point, not a law. The two mistakes I see most often are fixed day thresholds copied from a blog post and quintiles computed on a list that is 70% dormant, which compresses the top bucket into meaninglessness.
Anchor recency cutoffs to your buying cycle. If your average repurchase interval is 45 days, a customer at 90 days is already at risk and should not be in your “active” bucket. If your cycle is 18 months, a 90-day cutoff is noise. Pull the median inter-purchase interval from your own order data and set your first recency break at roughly 1.5× that figure, your second at 3×.
Two rules of thumb worth adopting:
- Never let any single segment exceed about 40% of the mailable list. If it does, your cutoffs are wrong, not your customers.
- Recompute boundaries monthly on a rolling 12-month window. Quintiles drift as your list grows; stale boundaries silently reclassify people.
For frequency, be honest that most lists are bottom-heavy. If 60% of your customers have exactly one purchase, quintiles will split that group arbitrarily. In that case use fixed buckets instead: 1 order, 2–3, 4–6, 7+. The buckets should reflect meaningful behavioral differences, not statistical symmetry.
The Segment Map: Where to Route Each Group
Here is a working map. Scores are 1–5 on each dimension; “H” means 4–5, “L” means 1–2, and “M” means 3.
| Segment | RFM profile | Share of list (typical) | Message strategy |
|---|---|---|---|
| Champions | HHH | 5–10% | Early access, loyalty perks, referral asks. No discounts. |
| Loyal | MHH / HMH | 10–15% | Cross-sell and bundle offers; upsell to higher tier. |
| Potential loyalists | HHM / HML | 10–15% | Second-purchase nudge, category education, small incentive. |
| New customers | HLL | 10–20% | Onboarding sequence, first-repeat-purchase offer. |
| Promising | MLL | 10–15% | Content-led nurture; test intent with a low-friction offer. |
| Need attention | MLH / MMH | 10% | Re-engagement with a specific reason to return. |
| At risk | LHH / LMH | 10–15% | Win-back with a real incentive, time-limited. |
| Can’t lose | LHL / LLH | 3–5% | Personal outreach or highest-value offer; protect revenue. |
| Hibernating | LLL / MLL | 15–25% | One last reactivation attempt, then suppress. |
Notice what is missing: a “lost” segment. Anyone with no engagement for 12 months and no purchase in 24 should leave the mailable list entirely, not sit in a bucket receiving quarterly “we miss you” blasts. That single suppression rule will do more for your sender reputation than any subject-line test.
Where to host the site behind the list
None of the tactics in this article matter if the infrastructure behind them is slow,
poorly configured, or capped at a mailbox size you outgrow in a month. If you are setting up
a site, a landing page, or a sending domain from scratch, Hostinger is a sensible starting point: entry plans are cheap enough to
experiment with, the control panel exposes DNS records for SPF, DKIM and DMARC without a support
ticket, and the VPS tier is there when you outgrow shared hosting. What matters more than the
headline price is that it scales predictably as the list grows.
Writing Different Messages for Different Cells
The map is worthless if every segment gets the same template with a different subject line. The offer, the proof, and the cadence should all change.
High-value, high-recency: sell access, not discounts
Champions and Loyal customers respond to status and convenience. Give them early access windows (48 hours before public launch), free shipping thresholds rather than percentage discounts, and referral programs with meaningful rewards. Discounting this group trains them to wait for sales and erodes margin. If you must discount, make it a loyalty-exclusive code and cap the redemption window at 72 hours.
New and promising: reduce friction, not price
A first-time buyer’s biggest barrier is uncertainty, not cost. Send a short sequence: a “how to get the most from your purchase” email at day 3, a category-education email at day 10, and a modest second-purchase incentive at day 21. Keep the incentive small. Deep discounts at this stage anchor the customer to your sale price forever.
At risk and can’t lose: be specific about what changed
“We miss you” is the laziest email in the industry. At-risk customers need a concrete reason to return: a new product in the category they used to buy, a price change, a service improvement, or a genuinely time-limited offer. For Can’t Lose customers, a human-signed email from an account manager outperforms any automation. If your list is too big for that, at least route them to a dedicated landing page rather than your homepage.
Hibernating: one honest attempt, then exit
Send a single reactivation email that asks directly whether the subscriber still wants to hear from you, with a one-click preference option. If there is no engagement within 30 days, suppress. This is uncomfortable because it shrinks your list. It also improves inbox placement for everyone who remains, and deliverability gains compound.
Automating the Scoring Pipeline
You do not need a data team to run this. A workable weekly process:
- Export transaction and engagement events from your ESP or warehouse into a single customer table, keyed on a stable ID.
- Compute recency (days since last qualifying event), frequency (count of qualifying events), and monetary (sum of spend) per customer.
- Assign quintile scores with
NTILE(5)in SQL, or a rank-and-bucket formula in a spreadsheet if your list is under about 20,000 records. - Map the three scores to the segment names using a lookup table.
- Push segment tags back to your ESP as custom fields so your automation builder can branch on them.
- Snapshot the scores monthly so you can track migration between segments. Segment movement is often more actionable than the segment itself: a Champion who drops to At Risk in one month is an emergency; a Promising customer who climbs to Loyal is a cross-sell opportunity.
Run the pipeline on a fixed day each week and never mid-campaign. Recomputing segments while a send is in flight causes duplicate or missed inclusions, and the debugging is miserable.
Common Failure Modes
Scoring on opens. Since Apple’s Mail Privacy Protection, a large share of opens are pre-fetched by the mail client. If your recency signal is open-based, your “active” bucket is inflated by people who never read anything. Use clicks and purchases.
Using lifetime monetary value only. As noted, this hides decline. A rolling 12-month window catches the customer who was valuable and is drifting.
Treating segments as permanent. They are not. A customer who buys twice in a week should move up immediately, not wait for the monthly refresh. Consider event-triggered promotion for high-value actions (a second purchase within 30 days should trigger a Loyal tag).
Ignoring margin. Monetary value should be profit, not revenue, where you can get it. A customer who only buys during 40%-off events is not a Champion even if the top-line number looks good.
Over-segmenting. Eleven segments is already a lot of creative to maintain. Start with four: Champions, New, At Risk, Hibernating. Add cells only when you have the content to support them.
Measuring Whether It Worked
Compare revenue per recipient (RPR) and click-through rate against a holdout group that receives your previous one-size-fits-all campaign. Run the test for at least four sends per segment; RFM segments vary in size and a single send is too noisy to trust. Expect the biggest gains in the At Risk and Hibernating cells, where targeted messaging replaces generic blasts, and the smallest gains in Champions, who were already converting.
Also watch list health: unsubscribe rate, spam complaint rate (keep it under 0.1%), and the share of the list that is mailable. If suppression is working, your mailable list should shrink in the first quarter and your engagement rates should climb. That trade is the point.
Related reading
- Email Segmentation Strategy: Beyond Demographics
- 12 Legitimate Ways to Grow an Email List That Actually Converts
Conclusion
RFM is not sophisticated, and that is its virtue. Three numbers, computed weekly from data you already own, will tell you more about what to send than most persona frameworks ever will. The discipline is in the details: score on clicks and purchases rather than opens, anchor recency cutoffs to your actual buying cycle, recompute boundaries monthly, and suppress the dormant instead of mailing them forever. Get those right and the segment map almost writes the campaigns for you.
Disclosure: this article contains an affiliate link. If you buy through it we may earn a commission at no extra cost to you. We only link to services we would recommend on merit.
Pingback: Moosend vs Klaviyo: Which Is Better for Newsletter Design Tools in 2026? – emailversus.com