Issuing, Acquiring and Wallets on One Platform. They Do Not Want the Same Thing.
One core, three business lines, three incompatible clocks. The constraint is never the platform — it is the missing mechanism for deciding whose requirement loses this quarter.
A processor that runs all three lines on one core looks efficient on an organisation chart. Underneath, it is three businesses with incompatible clocks sharing a single ledger.
There is a diagram that appears in the strategy deck of every regional payments processor. Three boxes across the top — issuing, acquiring, wallets — and one box underneath, labelled platform. The arrows all point down. The story is synergy: one core, one team, one investment, three revenue lines.
The diagram is not wrong. It is simply drawn from above, and almost everything that matters about running those three businesses together is invisible from that angle.
I want to write down what it looks like from underneath, because I have not seen it written down anywhere, and because the people who commission the diagram are rarely the people who live inside it.
Three businesses, three clocks
Start with what each line is actually optimising for, because they are not variations of one objective. They are different objectives that happen to involve cards.
Issuing optimises for the approval decision. Its unit of work is an authorization, its clock is measured in milliseconds, its enemy is a false decline, and its customer is a cardholder who will blame the bank for anything that happens at a checkout anywhere in the world. The issuing business lives and dies on a trade-off between approval rate and fraud loss, and that trade-off is adjusted continuously.
Acquiring optimises for settlement. Its unit of work is a batch, its clock is the daily cycle, its enemy is a funding delay, and its customer is a merchant whose business depends on money arriving when the contract says it will. The acquiring business lives on accuracy and predictability. Nobody has ever thanked an acquirer for being fast and approximately right.
Wallets optimise for balance. The unit of work is a stored value position, the clock never stops, the enemy is an incorrect balance shown to a user, and the customer is both a consumer and, increasingly, a business. A wallet is a deposit-like product wearing a payments costume, and its failure mode — a balance that is wrong, even briefly — is more corrosive to trust than a declined transaction.
Milliseconds, days, and continuous. Approval rate, settlement accuracy, and balance integrity. Three objectives, three time horizons, and one platform team with one release calendar.
Where the conflict actually shows up
The conflict is rarely a shouting match. It is a series of small decisions that each look reasonable and that together determine which business is being run well.
The release window. There is one. Every change to a shared core competes for it, and the competition is not between features. It is between a fraud rule that the issuing business needs before a holiday weekend and a settlement report change that the acquiring business promised a merchant two quarters ago. Both are correct. One of them moves.
What decides it, in most institutions I have seen, is not strategy. It is whoever has the better relationship with the person who owns the calendar. That is not corruption; it is the absence of a decision-making mechanism, and in its absence relationships become the mechanism.
The change that helps one line and quietly taxes another. This is the expensive one. A change to how authorizations are stored improves response times and makes a settlement report slower to produce. A new wallet balance model improves consistency and adds a step to the daily acquiring close. Neither effect is visible in the approval that authorised the change, because the approval was written by the line that benefits.
The cost appears two months later as an operations complaint, and by then nobody connects it to the change. I would guess that a material share of the chronic operational friction in any multi-line processor consists of taxes imposed by one business on another, invisibly, with everyone acting in good faith.
The incident. When the core has a bad night, all three businesses are affected and all three have different ideas about what recovery means. Issuing wants authorization restored immediately, even degraded. Acquiring wants the settlement cycle to complete correctly, even if that means later. Wallets want balances correct before anybody sees them, which may mean showing nothing for a while.
Those three preferences are in direct conflict during the exact hour when nobody has time to negotiate. If the order of recovery has not been agreed in writing beforehand, it gets decided by whoever is loudest on the call, and the institution discovers its real priorities at three in the morning.
The dispute where you are both sides. This is the one that surprises people who have not lived it.
A chargeback has an issuing side and an acquiring side. In a processor that runs both, there will eventually be a dispute in which the cardholder's issuer and the merchant's acquirer are the same organisation — sometimes the same team, occasionally the same person. Two of your clients, opposite sides, one set of hands.
The scheme rules anticipate this better than most institutions do, because the rules are written for a world where the two roles are separate and they do not stop applying when the roles collapse. What the rules cannot supply is the internal arrangement: who represents whom, who may not see which evidence, and who decides when the two internal positions disagree.
Most institutions discover they need that arrangement at the moment they need it. It should be written before the first case, and it is the kind of document nobody is assigned to write because it does not belong to any one business line.
The ledger is the actual product
Strip away the three boxes and what is shared is not a platform. It is a ledger.
Everything else — the authorization engine, the clearing files, the wallet balances, the dispute workflow, the reporting — is a way of writing to that ledger or reading from it. The ledger is where the three businesses are genuinely one business, and it is where a mistake in one line becomes a problem in another.
Which has a practical consequence I would put in front of any processor's board. The scarcest expertise in the organisation is not the people who understand the authorization flow, or the people who understand scheme mandates. It is the small number of people who understand how the ledger actually behaves — what is posted when, what is provisional, what is reversible, and why a balance that looks wrong at eleven at night is correct by six in the morning.
That knowledge is usually undocumented, usually held by fewer people than anyone is comfortable with, and never on a succession plan. It is the single largest operational risk in a multi-line processor and it does not appear on any risk register I have read, because risk registers record events and this is an absence.
What actually works
I am not going to pretend there is a clean answer, because the structure is genuinely hard and the alternative — three separate platforms — is worse in most of the ways that matter. But three things help, and all three are organisational rather than technical.
An arbitration function with a name on it. Somebody senior whose explicit job is to decide, in writing and with reasons, whose requirement loses this quarter. Not a committee. A person, with the authority to disappoint a business line and the obligation to explain why. The reasons matter more than the decisions, because reasons accumulate into a doctrine and doctrine is what lets the next decision be made without the meeting.
A cross-line impact statement on every change. One paragraph, written by the requesting line, naming what this change does to the other two. It will be wrong half the time. It is still transformative, because the act of writing it forces somebody to ask, and the half that is right would otherwise have been discovered as an incident.
A recovery order agreed in daylight. Write down, before you need it, which business is restored first and under what conditions. Have the three line owners sign it. Then it is a decision rather than an argument, and the argument happens when everyone is rested.
None of those requires a budget. All three are usually missing.
From a Mexico City desk
In México the same institution very often runs all three lines itself, which means the conflict I am describing is internal rather than contractual — and internal is harder.
When the issuing business and the acquiring business belong to different companies, they argue through a contract. There are service levels, there is a commercial relationship, there is somebody whose job is to enforce it, and the disagreement has a venue. When they belong to the same company, there is no contract and no venue. The disagreement is resolved by hierarchy, which means it is resolved by whoever is closer to the person who decides budgets, which has nothing to do with which business is right.
This is the structural reason I have come to believe internal service agreements are worth the effort people think they are not. Not the ceremonial kind that nobody reads — the two-page kind that says what one internal line owes another, by when, and what happens when it does not arrive. Their value is not enforcement. It is that they create the venue.
The second observation is about scale, and it is the one I would put to anyone building a regional processor. Scheme mandates do not scale down. A network's compliance deadline is the same date for an institution with four hundred million cards and one with four hundred thousand. The work is not proportional to the portfolio, and in a multi-line processor it arrives three times — once for issuing, once for acquiring, once for whatever the wallet is classified as.
Which means the fixed cost of simply remaining compliant is the most under-modelled number in regional payments. It is the real reason consolidation happens here, and it is almost never the reason anyone gives.
My judgment
The synergy in a multi-line processor is real, and it is not where the deck says it is.
It is not in the shared platform, which is a cost centre with three landlords. It is in the shared ledger and in the data on top of it — knowing both sides of a transaction, seeing the merchant and the cardholder, being able to answer a question that a single-line institution cannot answer at all. That is a genuine and defensible advantage, and almost nobody exploits it, because exploiting it requires the three businesses to agree on something and the organisation has no mechanism for that.
So my view, held firmly: the constraint on a multi-line processor is never the technology. It is the absence of a decision-making structure proportional to the number of businesses sharing the core. Institutions add a business line and do not add a way to adjudicate between business lines, and then spend years describing the result as a platform problem.
The second thing I believe, more quietly: the ledger people should be paid more and documented better than anyone currently thinks is necessary, and the test of whether an institution understands its own operation is whether it can explain its posting rules to somebody new in a week.
Three questions for the people who own this
1. The last time two business lines wanted the same release window, who decided — and was the reason written down anywhere that the next person could read?
2. If the core has a bad night, what is the agreed order of recovery across your lines? If the answer is "it depends on the incident," that is not an answer; it is a negotiation scheduled for the worst possible moment.
3. How many people in your organisation can explain, without looking, how and when a transaction posts to your ledger and what makes a posting reversible? Write the number down. If it is fewer than four, that number is your actual operational risk rating.
Reply to this email with a line or two. The sharpest answers will shape the next issue.
Sources
None. This essay cites no news event, no document and no figures, because it is about a structural pattern rather than an announcement. Everything in it is drawn from my own experience of running technology for a multi-line payments operation and from conversations with people who run others.
It is written as a pattern on purpose. There are no institutions named, no clients, no vendors, no platforms, no contract terms and no numbers from anywhere I have worked, and there will not be in anything I publish here. Where I say "in most institutions I have seen" or "I would guess," those are exactly what they claim to be — generalisations and estimates, offered so that a reader can disagree with them from their own experience rather than take them as findings. If your processor works differently, I would like to hear how. Nothing here reflects the position, portfolio, clients or vendors of my employer.