For banks
You bank them at the start, and lose them the moment they become worth banking.
Start-up propositions are measured in accounts opened and businesses supported. The commercial side is measured in balances and transactions. One is counted too early to mean anything, the other only changes after the customer has already scaled. By the time the account shows it, the lending, the foreign exchange and the corporate advice have gone elsewhere.
The timing problem, drawn
By the time the account moves, the customer has already chosen someone else.
A business builds commercial capability quarters before that shows up in balances, borrowing or transaction volume. Your systems read the second line. The first one is invisible to you.
That gap is the entire relationship opportunity. It is when a customer is deciding who handles the lending, the foreign exchange and the corporate advice that comes next, and it closes before the account tells you anything.
Measured twice a year, the first line becomes visible.
An illustration of the timing gap, not observed data. Both lines are indexed 0 to 100.
Two buyers, one instrument
Every bank has this problem twice.
Accelerators, hubs, partnerships
Runs on grant or marketing money and reports in participation. Needs per-company evidence that the programme changed something, before the next budget round asks for it. This buyer needs the quarterly report.
Business and commercial banking
Hundreds of accounts per manager, and a customer's growth is discovered from account behaviour after the event. This buyer needs the signal: which customers to spend scarce attention on, and what they need next. That’s where the value compounds.
The pilot starts on the programme side, because the evidence problem is funded and immediate. The relationship side is where it grows.
The firewall, stated first
It’s not a credit signal and must never be used as one.
The Growth Score does not assess creditworthiness, does not feed lending decisions, and is not an input to any affordability, pricing or risk model.
We state this before you ask and we put it in the contract. Your risk function doesn’t have to take that on trust either: the method, the arithmetic and the conformance vectors are published, so they can audit the instrument before a pilot rather than after one.
What every company in your cohort receives
A ten-page read: where the business stands, what it has built against what it is actually running, the stage and root cause holding it back, and the first three things to do about it.
Free to them, whether or not they ever speak to you again. The same result is what aggregates into your reporting, so there is one artefact and two audiences rather than a company view and a separate management view that can disagree.
What changes in the job
- Relationship managers get a reason to call. Not a product push, a specific finding about that customer's business
- Attention goes where it pays. Scarce RM time aimed at customers building capability, rather than spread evenly across a book
- The accelerator gets an answer for its board. Movement per member, quarterly, instead of attendance
- Churn is visible earlier. A customer outgrowing its banking shows in capability before it shows in the account
- The SME commitment becomes reportable. Something concrete for the annual report beyond lifetime totals
The payoff
What exists after one cycle that doesn’t exist now.
| Measure | Today | After one cycle |
|---|---|---|
| Which programme members are building commercial capability | Unknown until it shows up in the account | A dated position per member, refreshed each cycle |
| Evidence the programme produced anything | Attendance and satisfaction | Movement per company, baseline to endline |
| What a scaling customer needs next | Discovered in conversation, if at all | A named gap, recorded, ahead of the account showing it |
| The board answer on the SME commitment | Aggregate lifetime totals | A quarterly report, per cohort, comparable year on year |
Wearing your brand or ours. A bank-branded deployment runs the identical published method and declares its conformance at the same fixed path as every other, so co-branding never means a private variant.