Every lever, mapped to a number
Five headline KPIs make up the growth scorecard. Every segment is tied to the one it moves most, so strategy and measurement stay in step.
Where it sits: The Measure stage of the loop. A model that cannot name the number it moves is a classification exercise, not a programme.
The numbers you already report
Each one names the segments and the plays that move it, so a target has a mechanism behind it rather than a hope. These are the 5 on the Growth scorecard, in your language, not ours.
First-purchase conversion
Share of new registrations that complete a first purchase.
Active customers
Customers with a purchase inside the active window.
Purchase frequency
Average purchases per active customer per period.
Spend per active
Average spend per active customer per period.
Category breadth
Number of distinct categories purchased per customer.
Your framework covers 2 of these eight
A scorecard is a choice about what to watch, and it usually gets made before there is a model capable of moving much of it. These 8 are moved by RFM-BT in any vertical. 2 are already on the Growth scorecard above, which is a good sign: it means the framework was set with the right instincts. The other 6 move whether they are reported or not, and they are where the case for the model is usually won or lost. The percentage on each is the share of your base the named segments actually cover, from the same modelled distribution as the tuner.
Is the base worth more than it was?
Spend per active customer
Already on the Growth scorecardWhat an active customer is actually worth over a quarter, rather than an average across a base that is mostly dormant.
Reactivation rate
Not currently reportedThe share of quiet customers who come back, and crucially whether they stay back rather than making one purchase and going silent again.
Is what we have holding?
Lapse rate, against personal rhythm
Not currently reportedHow many customers are going quiet relative to their own pattern, which catches a weekly buyer at day ten instead of waiting for a fixed thirty-day rule.
Tier movement, up against down
Not currently reportedNet movement between value tiers over a quarter. The clearest single read on whether the base is improving, and the one a campaign report cannot produce.
Are they using more of us?
Products per customer
Already on the Growth scorecardHow much of the range a customer actually uses. Breadth is the strongest single predictor of whether someone is still here next year.
Complementary range take-up
Not currently reportedFirst-time adoption of the cross-sell target, and whether a second purchase follows the first. One conversion that never repeats is not adoption.
Is it costing less to get?
Spend per message sent
Not currently reportedWhat the programme earns per contact. This is the number that quietly falls when the answer to every target is to send more to everyone.
Opt-out rate by value tier
Not currently reportedWhether the cost of a busier programme is being paid by your best customers. A rising opt-out rate concentrated at the top is expensive in a way total volume hides.
A number is only a target if you could see it move
A measure driven by two per cent of the base will not produce a readable weekly result, however good the targeting is. That is not a reason to ignore it, it is a reason to watch it rather than steer on it, and to say which you are doing. Pick the measures whose segments are big enough to carry a read, and roll the rest up a level.
One customer, several numbers
A customer rarely sits under a single measure. Most segments move more than one at once, and a single steady behaviour can protect one number while gently lifting another. That is the argument for one shared model rather than a model per KPI: every number is being worked at the same time, from the same view of the customer.
Four questions, not eight metrics
The grouping is doing work. Growth, Retention, Breadth, Efficiency are the four questions a commercial reader asks, and a programme that moves only one of them is usually borrowing from another. Growth bought with a rising opt-out rate among your best customers is not growth, it is a loan.
That is RFM working on sample data. The service page has the rest: what it needs from you, how long it takes to build, what it pairs with, and how the lift gets proven.