Articles / Learning brief
The real cost of payments
Your notes
In simple terms / 01
What this means in plain language
Uses a hypothetical bank and transaction volumes to estimate the operating economics and potential profitability of domestic and cross-border payment services.
A payment service earns revenue from fees and foreign-exchange margins, but those figures are only the visible side of the economics. A bank also pays for scheme access, clearing, liquidity, correspondent services, fraud controls, compliance checks, technology, operations, investigations, and customer support. The useful question is therefore not simply how many payments are processed, but how much contribution remains after each direct and shared cost is assigned. A volume model helps teams compare products and see where automation or better routing could improve the result.
Complete lesson / 02
Understand the full idea, step by step
A payment receipt shows one fee. The payment itself consumed a dozen things that never appear on any receipt: a correspondent's charge, a slice of an FX spread, money parked in accounts so settlement never waits, the platform that processed it, and — for the unlucky few — an analyst's afternoon. What does a payment actually cost?
Where the cost of a payment arises
- Fees
- Explicit charges — the sending bank's fee, correspondent and scheme charges along the route
- FX
- The spread between the reference rate and the customer rate on any conversion
- Liquidity
- Money parked in nostro and settlement accounts so payments never wait for funding
- Operations
- The platform, screening, monitoring, and people that every payment consumes a share of
- Exceptions
- Repairs, investigations, and returns — rare per payment, expensive per case
The visible layer: fees and FX
Two of the five appear on Asha Traders' side of the deal, and the previous lesson measured them: Bank Alfa's transfer fee of EUR 15.00, and the FX spread worth EUR 216.34 against the reference rate. Under the SHA charge option a receiving-side deduction of USD 18.00 also came out of the amount — borne by the supplier, but still a cost of this payment. Everything the customer can see, in other words, can be added up from the quote and the charge option.
| Bank Alfa's transfer fee (paid by Asha Traders) | EUR 15.00 |
|---|---|
| FX spread versus the reference rate | EUR 216.34 |
| Receiving-side deduction under SHA | USD 18.00 |
| Deduction restated at the reference rate (18.00 ÷ 1.0800) | EUR 16.67 |
| All-in visible cost of this payment | EUR 248.01 |
This total takes the perspective of the deal as a whole — everything Asha Traders and its supplier together gave up compared with converting at the reference rate and paying no charges. It is an illustrative teaching total: real quotes shift with the market, and different routes carry different deductions. Notice what it excludes — every cost the bank absorbs to make the payment possible at all.
The hidden layer: liquidity, operations, exceptions
Bank Alfa keeps dollars sitting in its nostro account at Meridian Bank precisely so this payment could be funded on time — and idle money has a cost, because it could otherwise be earning elsewhere. The payment also consumed a slice of shared machinery: the processing platform, sanctions screening, fraud monitoring, reconciliation, and the staff behind them. Spread over millions of payments each slice is small, but it is never zero, and it must come from somewhere. Pricing that ignores the hidden layer looks generous right up until the product loses money.
You may be wondering: if the hidden costs are shared across millions of payments, why do they matter for any single one?
Because products differ in how much of the shared machinery they touch. A domestic transfer that validates cleanly and settles in a batch consumes almost nothing beyond its slice of the platform. A cross-border payment touches FX, correspondent accounts, heavier screening, and a much higher chance of manual work. Averaging everything together hides exactly the difference a pricing or routing decision needs to see — which is why Maya prices one payment at a time before averaging anything.
WHAT IF — A payment stops in the repair queue — a mistyped account number, an unclear beneficiary name, or a screening alert needing review
What happens: The payment still completes, but only after manual investigation. At an illustrative internal staff cost of EUR 30.00 per case, one exception can consume more than the fee income of dozens of clean payments.
How it is handled: Maya's team logs the reason code for every repair. Patterns matter more than cases: if one corridor or one customer's files keep needing repair, the economic fix is better validation at capture — preventing the exception — not faster handling of it.
COMMON CONFUSION
“Our payment product handles enormous volume, so it must be profitable.”
Volume multiplies whatever the per-payment economics are — including losses. A product earning EUR 1.00 per payment while consuming EUR 1.20 in platform share, liquidity, and exceptions loses more money the more it grows. Profitability is decided per payment and per product, by comparing all five cost sources against income; volume only scales the answer.
STRICTLY SPEAKING
Strictly speaking, every internal figure in this lesson — the EUR 30.00 exception cost, the platform slice, the liquidity charge — is illustrative. Real banks build unit-cost models with their own finance, treasury, and operations data, argue about how to allocate shared costs, and express uncertain inputs as ranges. The five cost sources are general; the numbers attached to them never are.
FOR NOW, REMEMBER
- Payment costs arise in five places: fees, FX, liquidity, operations, and exceptions.
- Fees and FX are visible to the customer and can be totalled from the quote — EUR 248.01 all-in for Asha Traders' illustrative payment.
- Liquidity, operations, and exceptions are borne by the bank, shared across payments, and consumed unevenly by different products.
- Volume never rescues bad per-payment economics; it only multiplies them.
TRY IT YOURSELF
Bank Alfa compares two products. Product A: fee EUR 1.00, fully automated, almost no manual work. Product B: fee EUR 12.00, but a meaningful share of its payments needs manual investigation at roughly EUR 30.00 per case, and it ties up correspondent liquidity. Which conclusion is soundest?
You have now seen both the visible and the hidden cost layers. The topic behind them goes deeper into charge options, conversion mechanics, and how quotes are built.
KEEP GOINGKey takeaways / 03
Three things to remember
- 01
Payment volume alone does not show profitability.
- 02
Costs span technology, operations, risk, liquidity, and networks.
- 03
Unit economics reveal where improvement work matters most.
Practical use cases / 04
Where you would use this
A product manager compares the margin of domestic and cross-border transfer products.
An operations lead estimates savings from automating repair and investigation work.
A finance partner tests whether a proposed customer fee covers end-to-end delivery costs.
Worked example / 05
Put the idea into a real situation
Illustrative example: a bank processes 100,000 transfers and collects an average fee of $1.20, creating $120,000 in revenue. Network, processing, compliance, operations, and allocated platform costs total $95,000, leaving $25,000 before other overheads. If manual repairs affect 4% of transfers, the team can model whether better validation would lower that $95,000 cost base enough to justify an implementation project alone.
Evidence & review / 07
Evidence & review
General cost model for domestic and cross-border credit transfers; not specific to any scheme or jurisdiction
What this brief simplifies: All monetary figures are illustrative. Shared-cost allocation is reduced to a per-payment slice; real unit-cost models involve allocation debates, ranges, and institution-specific data. Liquidity cost is described qualitatively rather than priced.
Sources for this brief2
- Market practice
Correspondent banking (final report) ↗ — CPMI, Bank for International Settlements · Costs and frictions in correspondent banking
Published in July 2016; its statistics cover 2011-2015 and are dated, but the definitions and arrangement types remain widely used.
- Simplified educational illustration
Payments Signal editorial teaching models — Payments Signal · Five-cost-source model; all internal figures illustrative
Used wherever diagrams, scenarios, figures, or example values are didactic constructions rather than sourced facts; every such use carries a simplifications disclosure. All people, companies, banks, and list entries in examples are fictional.