The failures below recur across industries, group sizes and jurisdictions. None of them is a product problem, which is why buying a product does not fix any of them. Each has a structural cause, and each is visible in advance to anyone willing to look.
One: treating payment failure as a payment problem
A payment does not arrive. The finance team contacts the bank, receives a reference number, waits, and the payment surfaces eleven days later or does not. The episode is filed as bad luck.
It is not bad luck, it is the correspondent chain. An international payment passes through a sequence of institutions, each applying its own screening, cut-off times and fee deductions. The originating bank determines only the first step. What determines the outcome is the depth of the correspondent network behind it and whether the institution will trace a stalled payment when asked.
The structural remedy is to select settlement relationships on that basis and to hold a second route for material flows. The operational remedy is to know, before a payment fails, who is called and what they are able to do. An institution that issues reference numbers instead of answers has told you something about every future payment.
Two: correspondent concentration disguised as diversification
A group holds four banking relationships and considers itself diversified. Three of them clear the same currency through the same correspondent institution. When that correspondent revises its risk appetite, three of the four relationships degrade simultaneously and the group discovers the shared dependency at the worst moment.
Correspondent routes are not published prominently and clients rarely ask. They should. The question is simple and every institution can answer it: for each currency the structure uses, which institution stands between you and the destination market?
Relationships can be independent while their settlement routes are not. Only the routes matter when something breaks.
Three: legal-to-banking drift in group structures
Entities are formed, acquired, made dormant and wound up. The accounts attached to them persist, because closing an account is nobody's project.
The result is a familiar set of symptoms. Cash sitting in subsidiaries that no longer trade. Guarantees outstanding against entities that have been restructured. Intra-group flows running through routes designed for a group configuration that no longer exists. Accounts in the name of entities whose directors resigned two years ago, discovered when the institution requests updated authority.
The correction is mechanical and unglamorous: a map of legal entities against banking relationships, a stated function for each pairing, and closure of what no longer serves one. It consistently produces the single largest improvement in treasury clarity of any exercise in this article, and it requires no new product.
Four: a treasury position that must be reconstructed
Ask a group what it holds and where, and observe what happens. If the answer requires a spreadsheet to be assembled from several portals over two days, the architecture has already failed, whatever the balances say.
The symptom set is consistent. Liquidity has to be assembled manually across entities and institutions before anyone can state a position. Reconciliation absorbs finance staff who should be forecasting. Funding pressure in one entity is invisible to the group until it becomes urgent.
None of that is fixed by a treasury management platform layered over a fragmented structure. A platform reports whatever the structure gives it, faster. The fix is separation of function: operating flow apart from reserve, custody apart from settlement, structural holdings apart from personal or operating accounts, so that a position can be read rather than built.
Five: foreign exchange by default
Where currency conversion happens at the moment of payment, at whatever rate the receiving institution applies, foreign exchange has stopped being a decision and become a consequence. Nobody chose the rate, the timing or the counterparty, and over a year the cumulative effect is material and invisible.
The structural cause is the absence of a treasury function holding balances in the currencies the business actually needs. When each payment is funded from whatever account has money in it, conversion is forced at the worst possible moment, which is the moment of urgency.
A designed structure holds working balances in its trading currencies, converts on a decision rather than on a deadline, and treats exposure as a position to be managed rather than a fee to be discovered afterwards.
Six: onboarding treated as paperwork
Clients arrive at onboarding assuming it is an administrative formality delegated to an assistant, and are surprised when it takes four months or fails.
Onboarding is the institution forming a view. The documents matter, but the coherence of the presentation matters more: an ownership chart that reconciles to the register, a business rationale that explains the payment flows the institution will actually observe, source of funds evidenced rather than asserted, and consistency across every document and every entity. Where a group has been onboarded at several institutions over several years, inconsistencies have usually accumulated, and it is the pattern rather than any single item that causes difficulty.
Preparation is straightforward and rarely done. Assemble the pack before approaching an institution, reconcile it internally, and be able to answer in one paragraph why this relationship is being opened and what will flow through it.
Six failures, six structural causes
- Payments fail because the correspondent chain was never assessed.
- Relationships correlate because settlement routes were never mapped.
- Accounts outlive entities because closure is nobody's responsibility.
- Positions are reconstructed because function was never separated.
- Currency conversion is forced because no treasury balance exists.
- Onboarding stalls because a view is being formed and no case was made.
The common thread
Every failure above is a structural condition presenting as an operational irritation. That is why they persist: the symptom is dealt with by whoever is nearest, repeatedly, and the cause is never examined because no single incident justifies the examination.
The test for whether a structure has these problems is not a question about banking at all. Ask how much of the finance function's month is spent explaining what already happened, and how much is spent deciding what happens next. The ratio is the diagnosis.