Customer segment migration: where the value flows
For your segments (for example from RFM), ordered from the best to the worst, it counts how many customers moved up, stayed or moved down a step over a period and what it cost in money. Shows where the base loses value and what one step down costs.
At a glance
- What question does it answer?
- Where the base loses value: from which segments customers move down and what it costs.
- What data is needed?
- Segments from the best to the worst (for example by RFM) and the number of customers at the start, moved up, stayed and moved down.
- How to read the result?
- On the example of four segments (21,000 customers) 1,140 moved up a step (5.4%), 2,140 moved down (10.2%), a net flow of −1,000 customers. The value of the base fell from 94.6m to 90.42m: the migration effect is −4.18m (moves down took 11.68m, moves up returned 7.5m). The most expensive moves down come from the segment “At risk”: 1,500 customers and 6.3m of annual margin. Example on the default values: Customers in the segments: 21,000; Moved up a step: 5.4% · 1,140; Moved down a step: 10.2% · 2,140; Net flow: moved up minus moved down: −1,000. The figures are illustrative: enter your own to see your result.
Why and how to reason
A segment (for example “Best”, “Loyal”, “At risk”, “Lost”) is a step on the customer value ladder. Customers move along it: rise, stay or fall. Money is lost not only when a customer leaves entirely but also when it drops a step. The calculation counts this directly: rise effect = risers × (value of the neighboring upper segment − own value), the fall effect is computed the same way and comes out negative. In the example annual customer value is 24,000, 12,000, 5,000, 800. From “Best” 180 fell: 180 × (12,000 − 24,000) = −2.16 million; from “Loyal” 460 fell: −3.22 million; from “At risk” 1,500 into “Lost”: −6.3 million. Falls cost 11.68 million, rises return 7.5 million, net −4.18 million: base value dropped from 94.6 to 90.4 million. The main loss is in the middle of the ladder, not at the very top.
When to use it
It comes in handy after RFM or another segmentation to understand where to direct retention: those falling need a reaction before they leave. It fits a brand or retailer with purchase history. It does not fit when segments are not ordered from best to worst or when you do not know the average value of a segment. Assumptions: annual customer value on a step is constant; migration is measured over one period; enter retention spending if you want to see the net effect.
How to read the result
Look at “net flow” and “migration effect”: if risers are fewer than fallers, the base loses value even if the number of customers does not change. Break the loss from falls down by step: where it is largest, that is where a campaign is needed. Compare retention spending with the effect: spending 2 million to stop an 11 million loss pays. But check that money goes where the fall begins (the “Loyal” and “At risk” steps) and not where the customer is already lost: winning back is dearer than retaining. Then look at the win-back calculation.
Common mistakes
Setting segment value by eye: then the money is nominal. Counting customers twice when the move happened within the period. Forgetting that a casual buyer can also “rise” and never come back. And one more: counting every fall as a loss although some customers return next period. Check migration over two periods in a row.
What data is needed?
- Segments from the best to the worst (for example by RFM) and the number of customers at the start, moved up, stayed and moved down.
- Margin per customer a year for each segment; optionally the retention spend.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Retention and reactivation spend for the period | ₽ | — | Optional. Mailings, bonuses and retention discounts. Needed to compare the migration effect with the spend. Amount excl. VAT: the general rate is 22% from 01.01.2026. The rate follows the shipment date; confirm your regime and rate with an accountant. |
How it is calculated
| Step | Formula |
|---|---|
| Effect of a move up | moved up × (value of the adjacent higher segment − own value) |
| Effect of a move down | moved down × (value of the adjacent lower segment − own value), a negative number |
| Check | value at the end − value at the start = migration effect |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Customers in the segments
- 21,000
- Moved up a step
- 5.4 % · 1,140
- Moved down a step
- 10.2 % · 2,140
- Net flow: moved up minus moved down
- −1,000
- Value of the base at the start
- 94.6 m ₽
- Value of the base at the end
- 90.4 m ₽
- Migration effect (moves up minus moves down)
- −4.18 m ₽The change in the annual margin of the base only from moves between adjacent segments.
- Loss from moves down
- −11.7 m ₽
- Effect net of spend
- enter the programme spend
Movement by segment
| Segment | At the start | Moved up | Stayed | Moved down | At the end | Share staying |
|---|---|---|---|---|---|---|
| Best | 1,000 | 0 | 820 | 180 | 1,060 | 82 % |
| Loyal | 3,000 | 240 | 2,300 | 460 | 2,780 | 76.7 % |
| At risk | 5,000 | 300 | 3,200 | 1,500 | 4,260 | 64 % |
| Lost | 12,000 | 600 | 11,400 | 0 | 12,900 | 95 % |
The most expensive moves down come from the segment “At risk”: 6.3 m ₽ of annual margin. Start retention work with it.
A customer is assumed to change a segment by no more than one step per period. If moves are wider, shorten the period. The customer value per year is your estimate; there is no single standard.
Nuances and limits
- Setting segments out of order of value: moves up and down get mixed up.
- Allowing jumps over several segments: the calculation counts adjacent steps only, shorten the period for wide moves.