PRODUCTS
Nine analyses. Nine models that run. Six markets. Central-bank and supervisor references, dated. All of them stop at PBT.
The nine models name it, and it is a different one in each product. In microcredit it is the freed hours that are never captured. In cards, the size of the issued base. In vehicle finance, affordability read on its own. In mortgage, PTI and turnaround time.
All four sit on almost every institution’s dashboard today. None of the four changes the P&L on its own.
A lever is not a benefit on its own. All nine models demand the same chain: decision, operating variable, P&L line. And all of them stop before tax, because that is where what management can redesign ends.
← The thesis these models carryTHE FINANCIAL PRODUCTS
THE RAIL
“The rail becomes implementation; the flow remains strategy.”
Open the payments model →A rail is the infrastructure money moves over: Pix in Brazil, SPEI in Mexico, Bre-B in Colombia, account-to-account transfers in Chile, interoperable wallets in Peru. Each country built its own, and each is faster, cheaper and more interoperable than the last.
Which is exactly why charging for access to the rail no longer sustains a price. The strategic unit is neither the transaction nor the rail: it is the flow — who pays whom, why the money moves, and how urgent delivery is. This model opens the section because it settles that distinction before anything is compared.
THE MAP
And underneath the flow sits the topology: the components of the network, on both rails, with the businesses that live on top and what it costs to enter each segment. The map is common; the position is yours.
Open the ecosystem map →Each row is a model, ordered by how long the money stays exposed: from the millisecond decision to the thirty-year one. Open the one you want to verify.
Product
the decision lasts milliseconds
The lever that moves the result
Customer behaviour
The decision that repeats
The authorisation, in milliseconds
The unit
Account and transaction
The metric that is watched and moves nothing
The size of the issued base
“Issuing more cards does not create value. Activating profitable behaviour does.”
Product
short flow
The lever that moves the result
Loan officer time capacity
The decision that repeats
Assess and originate in the field
The unit
Officer-hour
The metric that is watched and moves nothing
Freed hours
“If captured FTE = 0, this transformation produces no personnel saving.”
Product
36 to 55 months of exposure
The lever that moves the result
Exposure structure — LTV × term
The decision that repeats
Structure each approved loan
The unit
Loan-vehicle
The metric that is watched and moves nothing
Affordability, read alone
“It is diagnostic and does not change the P&L on its own.”
Product
20 to 30 years, with prepayment
The lever that moves the result
Price and duration
The decision that repeats
Underwriting and valuation
The unit
Contract-property
The metric that is watched and moves nothing
PTI and turnaround time
“TAT stays diagnostic until it changes conversion.”
The four metrics in the last column are the ones each model explicitly names as diagnostic: they inform, but they do not change the P&L on their own.
WHAT THE FOUR ROWS SAY TOGETHER
The metric that is watched measures activity. The lever that moves the result measures economics.
And the levers are of a different nature in each product — officer time, customer behaviour, exposure structure, contract duration. No single dashboard can govern all four.
A board reviewing four products with one set of indicators is measuring four different economies with the instrument of one. That is why there are four models and not one.
All four are decision models, not projections. They run across six markets with central-bank and supervisor references dated August 2026, keep those references separate from the editable illustrative assumptions, and stop at PBT: no tax, no PAT, no ROE.
ECONOMICS AND ARCHITECTURE
All three open by naming a management error and correct it with a model that runs. Each one also declares what it does not incorporate.
APIs
The error it corrects
Confusing markup with margin. A 40% markup on cost leaves 28.6% real gross margin.
What becomes money
Reserved capacity, priced in two parts: access plus per-call overage.
The instrument
Twelve inputs. Returns the tariff, the economic floor per overage call, and sensitivity at 60%, 100% and 140% of projected volume.
The verdict
“A price that only works at 100% of projected volume is not a robust price.”
INFORMATION
The error it corrects
Adjusting data quality until it yields the desired price. “The quality model became a justification.”
What becomes money
Not the data: the decision the data changes — approve, price, prioritise, detect, retain.
The instrument
Twelve inputs. Returns a corridor between the cost floor and the value ceiling, and where the price falls inside it.
The verdict
“If Pmax < Pmin, there is no sustainable commercial case.” The model can return: do not commercialise.
OBSOLESCENCE
The error it corrects
Deciding by age and calendar instead of by the cost of keeping.
What becomes money
There is no price here: there is avoided cost. Keeping is a capital allocation decision.
The instrument
An S-curve with five inputs. Returns time to target, maximum speed in points per month, and what the last point costs.
The verdict
“If the cost of keeping is not measured, ‘doing nothing’ will almost always look cheaper than it really is.”
THE DOOR
Clients and products per employee is the only one of the nine that does not measure a product: it measures how much structure it takes to run them. That is why it is the door to the agents.
Leverage gap = growth in economic output − growth in operating structure
The distance between the two curves is the only measure that says whether something changed the economics or only changed the plumbing. When they run parallel, volume grew and economics did not.
An agent does not change a product’s economics directly. It changes how much structure it takes to run it. This gap is, literally, the variable an agent moves — and the same one that carries the Strategy thesis.
WHAT THE MODEL BRINGS
Five metrics in a ladder: scale density, load density, economic productivity, cost to serve and a quality guardrail.
Five comparability rules, because a bad denominator produces false productivity.
The explicit separation between what the filings support and what management still has to prove.
Public FY2025 evidence on Nu, Revolut, SoFi, Santander and Interactive Brokers.
ONE METRIC, FOUR AGENTS
Each agent moves a different slice of the same gap.
PMO / PM
The execution structure: how many people it takes to move the change portfolio.
CPO
The product structure: how many people it takes to run the products on this page.
CFO
The finance structure: close, reconciliation and control.
CLO
The contract structure: origination and contract administration.
PMO is downloadable today. CFO, CLO and CPO are coming.
