Docs · Reports
Customer segments report
The customer segments report (RFM: recency, frequency, monetary) places every customer in one behavioural group, shows where the revenue sits and who is slipping, tracks which months’ customers came back, and turns all of it into ranked, testable hypotheses. This page walks through every section.
Before you start
Start one from your workspace with New Analysis and pick Who your customers are. Quick run runs right away with sensible presets. Customize adds three optional steps:
- Segmentation
- Default, or one of your custom segmentations (see below).
- Customers to analyse
- All customers, or only those whose first order was in the last 90 days.
- Compare against
- Previous 30, 60, 90 or 180 days. Segment changes and the migration section compare today with where each customer stood that many days before the last order in your data.
- Schedule
- Re-run the analysis automatically on fresh data.
Good to know
The report needs a Customer ID or email column. Orders without one are left out. If your import had no customer column, re-import with it mapped.
Prices are optional. Without them, Monetary ranks customers by units bought instead of spend, and the report says so at the top.
How segments are defined
Each customer gets three scores from 1 to 5:
- Recency (R)
- Days since their last order, ranked against the rest of your customers. The most recent fifth scores 5.
- Frequency (F)
- Their number of orders, capped at 5: one order scores 1, five or more score 5.
- Monetary (M)
- Their total spend, ranked into fifths. Shown on every customer; the default segments do not use it.
The default segments are checked top to bottom, and a customer lands in the first one they match:
- Champions
- R 4 or 5 and F 4 or 5. Bought recently, buy often and spend the most.
- Loyal Customers
- R 3+ and F 3+.
- New Customers
- R 4+ with exactly one order.
- Potential Loyalists
- R 3+ (recent, not yet frequent).
- At Risk
- R 2 or lower and F 3+: used to buy often, have gone quiet.
- Hibernating
- R 2 or lower and F 2 or lower.
- Others
- Anyone left. Hidden when empty.
Segments are grouped into health buckets that colour the whole report: Secure (healthy revenue you expect to keep), Neutral (watched, not acted on), At risk (slipping, still cheap to bring back) and Drifting (gone quiet: win back or write off). Teal is healthy, wheat is at risk, terracotta is hibernating.
Section 1
The headline numbers
The first paragraph names where the revenue is: concentrated in one segment, spread across the base, or sitting in a segment that has gone quiet. The second describes the average customer and how quickly new customers come back: the share who order again in their first month, by month three, and what an acquired customer is worth after six months.
The revenue health bar splits lifetime customer revenue by segment, grouped into Secure, Neutral and Drifting. The lines under it say what each block is worth and how many customers hold it.
How to read it
The drifting share is the headline risk: revenue from customers who used to buy and have stopped. A large Hibernating block with a small At Risk block means most of the slippage already happened; act on At Risk first, while they are still cheap to bring back.
Hypotheses
Under the headline, the report ranks up to five testable claims, one per target segment, each with a type (Win-back, Nurture, Upsell, Retention, Churn prevention), a confidence level and a timeline. Examples: “Win back Hibernating before their $2.0M is gone”, “The second order happens by month 4 or not at all”.
Open one for its test plan: four concrete steps, the metric to watch and how long to run it.
The hypotheses come from fixed rules over your numbers, not from a language model, so the same data always gives the same list.
What to do with it
Pick one, create a space for it with the hypothesis as its claim, run the test, and record the verdict there.
Section 2 · Who they are
Which segments do your customers fall into?
- Customers
- How many customers are in the segment today.
- vs prev
- Change against the segment's size at the comparison cutoff. For At Risk and Hibernating the colours flip: growth there shows in red.
- Revenue, Share
- Lifetime revenue of the segment's customers and its share of the total.
Click a segment to open its profile:
- Customers and revenue, with the change since the comparison window.
- An observation and a recommended action, written from the segment’s own numbers.
- Average recency, frequency and value.
- The segment’s top selling products: what to feature when you write to them.
View in explorer filters the customer table below to this segment.
Section 3 · Where the money is
Which segments carry the revenue?
Tile size is the segment’s share of revenue. Hover a tile for the exact revenue and customer share. A small segment with a big tile is where a lost customer costs the most.
Section 4 · Who's moving
Who got better, who got worse?
Each band is a group of customers flowing from their segment at the comparison cutoff (left) into today’s (right). Teal bands moved healthier, terracotta bands slipped. The counters above sum it up: Moved healthier, Slipped, Stayed put and New this window (customers with no orders before the cutoff).
How to read it
The comparison window is the one chosen at creation (30 days by default). A short window shows recent movement; a longer one smooths seasonal noise. The section appears once customers have orders on both sides of the cutoff, and only in your dashboard, not on shared links.
Section 5 · Who comes back
Which months' customers stuck around?
Customers are grouped by the month of their first order (rows) and tracked month by month since (columns). Read down a column to compare cohorts at the same age; reading across a row mixes ages. Hatched cells have not happened yet.
- Retention
- The share of the cohort that ordered in that month. Per month, not cumulative.
- Revenue per customer
- That month's revenue divided by the cohort size.
- Cumulative value
- The running total of revenue per customer: what an acquired customer is worth by month N.
- Today
- A bar of each cohort's current segment mix. Hover for the breakdown.
- All cohorts
- The size-weighted average row, and the counters above the grid (Back in month 1, 3, 6, 12; Value at month 6).
What to do with it
If most returning customers come back by a given month (the line under the grid says which), time your second-order email before it. A recent cohort that falls below older ones at the same age points at a change in acquisition or onboarding.
The grid needs at least three cohorts with three months of history, and it is computed when Customers to analyse is All customers. For retention by the product bought first, see the products table in the market basket report.
Section 6 · Find a customer
Every customer, scored
Search by customer ID or filter by current and previous segment. Changes Only keeps customers whose segment moved. Click a customer for their R, F and M scores, segment history and last purchase date. This section is visible only in your dashboard.
Exporting customers
Export downloads the filtered list as a CSV. Marketing-ready formats match the import columns of Klaviyo, Omnisend, MailerLite, Mailchimp and HubSpot; Customize Columns lets you pick the fields and file name. Upload the file to your email tool as a list or segment. If your customer IDs are not email addresses, the export warns you first.
Custom segmentations
When the default segments do not fit your business, build your own in the Segmentation step of the create dialog. A segmentation is a list of rules checked top to bottom; a customer lands in the first rule they match, and a fallback catches everyone else.
- Conditions can use days since last order, orders, lifetime value, days since first order, average order value, days between orders, first order value, first and last order date, and the R, F and M scores.
- Each rule gets a name and a health bucket, so the report colours it correctly.
- A live preview shows how many customers from the latest run land in each rule.
- Saved segmentations belong to the workspace, keep a version number, and can be reused by every later analysis, including the segment overlay in market basket reports.
If a segmentation changes after a report ran, the report says so and asks for a re-run.