A 15%-off email goes out to the whole customer list on a Tuesday. One recipient is your best customer, who buys every month and was going to order on Friday at full price; you have just given away margin on a purchase you already had. Another is a first-time buyer who is still unsure about you and needed reassurance more than a discount. A third used to spend a small fortune with you and has not ordered in five months; for them, 15% off is too little, too late, and too impersonal to register. Same email, three wrong answers.
That is the case for value-based segmentation, and it starts from an uncomfortable idea: treating every customer the same isn't fair, it's expensive. Customers are worth very different amounts to your business and are heading in different directions. Recognising that, and treating each group differently on purpose, is one of the highest-return things a retention team can do with data it already has.
This post groups customers by value and direction. It sits alongside email segmentation, which covers segment types more broadly, and cohort analysis, which groups customers by when they joined.
Why value is worth segmenting on
In most businesses, value is concentrated. A frequently observed pattern is that a minority of customers, often somewhere around a fifth to a third, produce the large majority of revenue, and an even smaller group produces a large share of profit. The exact split varies widely by category, so check your own numbers, but the direction is almost universal: a small group matters far more than its size suggests.
That concentration changes how you should spend attention and money. A single well-treated group of best customers, and a single well-handled group of valuable customers who are starting to drift, can move results more than any campaign sent to everyone. It also changes what "good" looks like: the aim is not to squeeze more from every customer equally, but to move people up, keep the best ones close, and catch the valuable ones before they go.
Scoring customers with RFM
The simplest and most widely used method is RFM: recency (how long since the last purchase), frequency (how many purchases) and monetary value (how much they have spent). It needs nothing more than a list of orders, and it can be built in a spreadsheet.
Rank every customer on each of the three measures and split them into five equal groups, scoring 1 to 5, with 5 as best. The most recent buyers get a recency score of 5; the most frequent get a frequency score of 5; the biggest spenders get a monetary score of 5. Each customer ends up with a three-digit profile, such as 5-5-5 for a recent, frequent, high-value customer, or 1-5-5 for someone who used to buy often and spend well but has gone quiet.
Three refinements make the scores more useful. First, recency is usually the most telling signal of who is drifting, so watch it most closely. Second, where you can, use gross margin rather than revenue for the monetary score, so heavy discount users don't look more valuable than they are. Third, judge recency against your own buying cycle: sixty days of silence means very little for an annual purchase and a great deal for a weekly one. Clean data matters too, since duplicate records split one customer's history in two, and a single view of each customer, as argued in CRM consolidation, makes the whole exercise far more reliable.
Kinsta
Premium managed WordPress hosting with industry-leading speed and reliability
Best for: Premium WordPress Hosting
Five segments are enough
RFM can produce dozens of combinations, and some frameworks use eleven or more named segments. Most teams get more value from a handful they will actually act on.
| Segment | Who they are | What they need |
|---|---|---|
| Best | Recent, frequent, high value | Recognition and access, not discounts |
| Growing | Recent, still building frequency | Guidance and a reason to come back |
| Steady | Regular, mid-value | Consistency and gentle expansion |
| At risk | Valuable history, recency slipping | Fast, personal attention |
| Drifted | Low value, long inactive | One good attempt, then less contact |
These segments work alongside interest and behaviour, not instead of them. A best customer who loves one product range still wants to hear about that range; value-based segments decide how much attention and investment someone gets, while other segmentation shapes what the message is about.
The principle Treating every customer the same isn't fair, it's expensive. You discount people who would have paid full price and under-serve the ones about to leave. Different customers are worth different amounts and are heading in different directions; treat them that way on purpose.
Your best customers: recognise, don't discount
Your best customers are already buying often and spending well. Blanket discounts mostly give away margin on purchases they would have made anyway. What they tend to value more is recognition: early access to launches and sales, exclusive products, faster or more personal service, and a genuine thank-you now and then. Those cost little and signal that you have noticed them, which is exactly what a well-designed loyalty tier is for.
They are also your most credible advocates, which makes them natural candidates for referral programmes, reviews and product feedback. And protect them from over-mailing: the most engaged customers often receive the most messages, and fatigue can push even loyal people away.
Growing customers: build the habit
Growing customers are recent buyers who haven't yet settled into a pattern. They are the pipeline for your best segment, and they need help more than incentives: reassurance that they chose well, guidance on getting more from what they bought, and a relevant reason to return, such as a complementary product or a timely reorder reminder.
The biggest lever here is usually the second purchase, which is why this segment overlaps closely with the work described in customer onboarding after the sale. Steady customers need a lighter version of the same: consistency, useful content and occasional nudges towards products they haven't tried yet.
At-risk customers: the best return on attention
Customers with a valuable history whose recency is slipping are often the single best place to spend retention effort. They have already shown they like you, and they are worth a lot, but they haven't gone yet. Spot them early by watching for recency dropping well beyond their usual buying rhythm, rather than waiting for a fixed number of days, the same behaviour-based logic used in behaviour-based scoring.
Your at-risk group has 200 customers who each used to spend around $400 a year → $80,000 of annual revenue is drifting.
→ A personal outreach, such as a message from a real person, a phone call for the very top customers, or a considered gesture, costs roughly $20 each → $4,000 in total.
→ If it brings back just one in four of them, that recovers about $20,000 a year.
Five times the cost, from a group most businesses treat exactly like everyone else. Compare that with a blanket discount to the whole list, which spends margin on best customers who didn't need it and still feels impersonal to the at-risk ones.
Find out why before you act, using the approach in churn diagnosis, since a valuable customer who had a bad experience needs an apology and a fix, not a coupon. For the drifted, low-value segment, one sensible reactivation attempt through re-engagement automation is usually enough before reducing contact.
Keep segments moving
Segments are only useful if they reflect what customers are doing now. Recalculate scores at least monthly, and more often if your platform can automate it, because people move constantly: new buyers become regulars, regulars become best customers, and best customers start to drift. A segment built once and left alone quickly sends the wrong message to the wrong person, such as a win-back offer to someone who bought last week.
Most importantly, track movement between segments, not just their size. How many growing customers became best customers this month? How many best customers slipped into at-risk? Those flows tell you far more about the health of your customer base than any single snapshot, and they are the numbers worth reviewing in a regular funnel audit. Store segment history against each customer too; it is some of the most valuable first-party data you own.
The short version
Treating every customer the same isn't fair, it's expensive: blanket offers give away margin on your best customers' planned purchases and still fail to reach the valuable ones who are drifting away. Value is usually concentrated, with a minority of customers producing most of the revenue, so segmenting by value and direction pays off quickly. Score customers with RFM, using recency, frequency and monetary value ranked into five groups each, and refine it by watching recency closely, using margin rather than revenue where possible, and judging recency against your own buying cycle. Keep to about five segments: best, growing, steady, at risk and drifted. Treat best customers with recognition, early access and service rather than discounts, and protect them from over-mailing. Help growing customers reach their second purchase. Give valuable at-risk customers fast, personal attention after finding out why they are slipping, since that group often returns several times what it costs. Make one sensible attempt with drifted customers, then reduce contact. Recalculate segments regularly and track how customers move between them.
Are you treating your best customers like everyone else?
We help brands segment customers by value and build retention that fits each group.
Explore E-commerce Marketing →Frequently asked questions
What is value-based customer segmentation?
Value-based customer segmentation groups customers by how much they are worth to the business and which direction they are heading, rather than by who they are. Instead of dividing people by age, location or interest, it looks at purchase behaviour: how recently they bought, how often they buy and how much they spend or contribute in margin. That produces a small number of practical groups, such as your best customers, customers who are growing, steady regulars, valuable customers who are starting to lapse, and low-value customers who have drifted away. Each group needs a different kind of attention. The reason it matters is that value is usually concentrated, with a relatively small share of customers producing a large share of revenue, so treating everyone the same tends to over-discount the best customers and under-serve the ones who are quietly leaving. It complements other segmentation rather than replacing it, since the same valuable customer may still need messages matched to their interests.
How does RFM analysis work?
RFM analysis scores each customer on three measures from their purchase history: recency, meaning how long since their last purchase; frequency, meaning how many times they have bought; and monetary value, meaning how much they have spent. The most common approach ranks customers on each measure and splits them into five equal groups, giving a score from 1 to 5, where 5 is best, so the most recent buyers get a recency score of 5. Combining the three scores gives each customer a profile such as 5-5-5 for your best customers or 1-5-5 for someone who used to buy often and spend a lot but has not purchased in a long time. You can do this in a spreadsheet with nothing more than a list of orders. Recency is usually the most useful single signal for spotting who is drifting away, and it is worth using gross margin rather than revenue for the monetary score where you can, so heavily discounted buyers do not look more valuable than they are.
How should you treat your best customers?
With recognition, access and service rather than discounts. Your best customers are, by definition, already buying often and spending well, so blanket discounts mostly give away margin on purchases they would have made anyway. What they tend to value more is being recognised and looked after: early access to new products or sales, exclusive items, faster or more personal service, thoughtful surprises and genuine thanks. They are also often your most credible advocates, which makes them natural candidates for referral programmes, reviews and product feedback. Protect them from over-mailing too, since the customers who engage most are sometimes the ones who receive the most messages, and fatigue can push even loyal customers away. Watch their recency closely: a best customer who suddenly goes quiet is one of the most valuable people in your database to reach quickly, ideally with a personal message rather than an automated offer.
How often should customer segments be updated?
Often enough that a customer's segment reflects what they are doing now. For most online businesses that means recalculating scores at least monthly, and weekly or daily if your platform can do it automatically, because customers move between segments all the time: a new buyer becomes a regular, a regular becomes a best customer, and a best customer starts to drift. Static segments built once and left alone quickly become misleading and lead to the wrong treatment, such as sending a win-back offer to someone who bought last week. Beyond keeping segments fresh, track movement between them as a measure in its own right. How many growing customers became best customers this month, and how many best customers slipped into the at-risk group, tells you far more about the health of your customer base than the size of any single segment at one point in time.
