Skip to main content
Every night, Gameball’s RFM Segments place each of your customers into one of seven groups, such as Best Buyers or About to Lose. This article explains how each customer’s scores are worked out and how those scores decide their segment, with examples you can follow step by step.

​
How RFM Scoring Works

Each customer’s segment answers three questions. Each question gets a score from 1 (weakest) to 4 (strongest), and the three scores together decide the segment.

Recency

When did they last buy?Based on their latest completed order.

Frequency

How often do they buy?Based on their number of completed orders.

Monetary

How much do they spend?Based on their total completed spend.
Gameball calculates the scores in four steps:

1. Measure the Buying Cycle

Learn how often your returning customers usually come back.

2. Score Recency

Count how many buying cycles have passed since the last purchase.

3. Score Orders and Spend

Compare each buyer’s orders and spend with your other buyers.

4. Assign the Segment

Combine the scores to place the customer in one of seven segments.

​
Step 1: Measure the Buying Cycle

“Recent” means different things in different businesses. A grocery customer who hasn’t ordered in a month is drifting away, while a furniture customer who hasn’t ordered in a month is perfectly normal. Before scoring anyone, Gameball measures how often your returning customers usually come back. This is your store’s buying cycle.
1

Find your returning customers

Gameball looks at every customer who bought on two or more different days in the last two years.
2

Work out each customer's average gap

For each returning customer, Gameball calculates the average number of days between their purchases.
3

Take the middle value

The gaps are sorted from shortest to longest, and the middle value becomes your store’s buying cycle.
Sorted, the gaps are 10, 31, 60, 89, and 167. The middle value is 60, so this store’s buying cycle is 60 days: a typical returning customer buys about every two months.
Same-day orders count once: Two orders on the same day count as one buying day. A customer who splits a basket into two orders hasn’t really “come back”, so counting days keeps your buying cycle accurate.

​
Buying Cycle Limits

7 days minimum

A week is about the fastest a marketing team can act. Shorter windows would move large numbers of customers between segments every night.

180 days maximum

Beyond six months, “recent” stops meaning recent.

41 days for new stores

Used until your store has at least 100 returning customers and enough history to see the pattern repeat four times.
41 days is the typical buying cycle across all Gameball stores. Your buying cycle is re-measured every night, so it adjusts as your customers’ habits change.

​
Typical Buying Cycles by Industry

​
Step 2: Score Recency

Recency is counted in buying cycles. Using the 60-day example store:
Customers who haven’t bought yet are timed from the day they signed up. Within one buying cycle of signing up, they score 4 because they haven’t had time to buy yet. After one cycle with no purchase, they score 1.The signup date is the one your store shares with Gameball. If your store doesn’t share one, Gameball uses the day the customer was added to Gameball.

​
Step 3: Score Orders and Spend

For orders and spend, there is no “normal schedule” to compare against, so each buyer is compared with your store’s other buyers. Imagine every buyer standing in a line from lowest to highest, then cutting the line into four equal parts. Orders and spend are each scored this way.
If a business account that spent 968,654 SAR is added, it joins Nadia and Layla in the top quarter. Reem and Sami still score 1, and nobody’s score changes because of it.
Ties share a score: In most stores, many buyers have exactly one order, and they all get the same order score. This means the four order groups are rarely equal in size, and in some stores one score has nobody in it.

​
How Unusual Customers Are Handled

Scores are based on position in the line, not the amount. A business account with thousands of orders simply lands at the top and scores 4, like the rest of the top quarter. It can’t push anyone else’s score down.
Refunds for orders Gameball never received can leave a negative total. These are counted as 0, so a refund never ranks someone below a customer who hasn’t bought.
Customers with no completed order are treated as never having bought: they score 1 for orders and spend, and their recency is timed from their signup date.
Dates such as 1 Jan 1900 are ignored. Any date before the year 2000 counts as no purchase date, so it can’t make a customer look decades overdue.

​
Combining Orders and Spend Into a Value Level

The order score and spend score are averaged (rounded down) into one value level:

​
Step 4: Assign the Segment

Recency decides the row. For customers who are on schedule or have gone quiet, the value level decides the segment as well.
A customer who has gone quiet but used to buy and spend well is worth winning back, so they go to About to Lose instead of Lost.

​
What Each Segment Means

Best Buyers

On schedule, and in the top quarter for both orders and spend.Suggested action: Reward and recognize them with early access and VIP perks.

Loyal

On schedule, with solid but not top value.Suggested action: Nudge them up with bundles, tier goals, and referral offers.

New Customers

Bought within the last cycle with few orders and low spend, or signed up within the last cycle and haven’t bought yet.Suggested action: Send a welcome offer for the first purchase, then make the second one easy.

Promising

Missed one normal visit, at any value level.Suggested action: Send a timely reminder before the habit breaks.

Can't Lose

Two to four cycles without buying, at any value level.Suggested action: Run a re-engagement campaign with a reason to return.

About to Lose

Gone quiet, but used to order and spend well.Suggested action: Make a strong win-back offer. These are the most valuable customers to recover.

Lost

Gone quiet with low value, or signed up more than one cycle ago and never bought.Suggested action: Use low-cost reactivation, or a first-purchase offer for those who never bought.

​
Worked Example: Eight Customers in One Store

This store has a 60-day buying cycle. Expand each customer to see why they land in their segment today.
Bought 10 days ago · 14 orders · 2,400 SAROn schedule (10 ≤ 60 days), and in the top quarter for both orders and spend.
Bought 25 days ago · 3 orders · 410 SAROn schedule, with upper-middle value.
Bought 5 days ago · 1 order · 90 SAROn schedule, but one small order so far gives him a low value level.
Signed up 12 days ago · no order yetShe has had less than one cycle (12 ≤ 60 days) to make a first purchase, so she counts as new, not lost.
Bought 95 days ago · 2 orders · 120 SAR95 days is past one cycle but within two, so she missed one normal visit.
Bought 200 days ago · 6 orders · 300 SARBetween two and four cycles without buying. His value level doesn’t change this segment.
Bought 400 days ago · 9 orders · 900 SARGone quiet (past 240 days), but she used to buy often and spend well.
Bought 400 days ago · 1 order · 55 SARGone quiet, with low value.
Segments change as time passes: If Huda buys again tomorrow, she moves back to an on-schedule segment the next night. If Layla stops buying, she moves to Promising after 60 days, even though she hasn’t done anything. If Yara still hasn’t bought after 60 days, she moves to Lost.

​
Reading the Segment Summary on Your Dashboard

Each segment on the RFM Segments page shows a summary like this example from a store with a 72-day buying cycle:

Best Buyers

Bought recently and are in your top 25% for both number of orders and total spend. Last purchased 0 to 72 days ago, with an average of 7.4 orders and 1,758 SAR spent.
  • Days: The most and least recent purchase in the segment. For on-schedule segments, this is at most one buying cycle.
  • Orders and spend: The segment’s average. An average describes the typical member, while a lowest-to-highest range would mostly reflect one or two unusual accounts.
The summary numbers only count customers who have bought. New Customers and Lost also include customers who haven’t bought yet, but they are left out of these numbers so they don’t pull the averages down.

​
Common Questions

Because time passed. Recency is counted from today, so a customer who stops buying moves from on schedule to Promising, then Can’t Lose, then About to Lose or Lost. Segments update every night.
Yes. These two segments depend only on how long it has been since the last purchase. A big spender who hasn’t bought in two cycles is in Can’t Lose, which is exactly the type of customer you want to win back.
For one buying cycle after signing up, they are in New Customers because they haven’t had time to buy yet. If they still haven’t bought after one cycle, they move to Lost.
No. Only completed orders count toward orders and spend. A customer whose only order was cancelled counts as never having bought.
Most stores have many customers who bought once a long time ago, plus sign-ups who never bought. Measuring against your store’s real buying rhythm shows how many customers have actually gone quiet. About to Lose is usually your best win-back opportunity.
No. Only active customers on your live account are scored. Test, deactivated, and deleted customers have no RFM segment.
Segments are recalculated every night, so a new purchase moves the customer at the next nightly update.

Introduction to RFM Segments

Understand the concept of RFM analysis and where to find your segments.

Using RFM Segments to Serve Your Goals

Learn how to act on each segment with campaigns and retention strategies.