Two companies of similar size, trading in similar markets, each hold six banking relationships. On paper they look alike. In practice one of them can state its consolidated liquidity position in a morning and the other needs a fortnight and a spreadsheet nobody trusts.
The difference is not the number of accounts. It is whether the structure was designed or whether it merely arrived.
How structures arrive
Fragmentation has a recognisable life cycle, and it never begins with a bad decision. It begins with a series of reasonable ones.
A supplier in a second market wants payment in local currency, so an account is opened there. A subsidiary is incorporated and its auditor insists on a local banking relationship. A counterparty declines the group's existing bank as issuer of a guarantee, so a third institution is engaged for that transaction and then retained because closing it seems like effort for no benefit. A relationship manager leaves and the successor is slower, so a parallel account is opened at a competitor while the situation is assessed. The situation is never reassessed.
Five years later the company holds seven relationships across four jurisdictions, of which two carry meaningful balances, one is dormant but still generating periodic due diligence requests, and none has a written statement of what it is for.
No individual step was wrong. The aggregate is indefensible.
Why it is difficult to see from the inside
Fragmentation is invisible to the people closest to it for three reasons.
The first is that each relationship has an origin story. Someone remembers why the account was opened, and the memory of the original purpose stands in for evidence that the purpose still exists. Justification by history is the most common defence of a redundant relationship.
The second is that the cost is distributed. No single account is expensive. The burden appears as a general sense that finance is understaffed, that month end takes too long, that compliance requests keep arriving. None of those symptoms points at any particular relationship, so none is closed.
The third is that fragmentation is rewarded in the short term. Opening an account solves today's problem in days. Redesigning a structure solves next year's problem in months. Under pressure, the fast answer wins, and it goes on winning until the structure becomes the problem.
What a designed architecture looks like
An architecture assigns function. Every relationship exists for a stated reason: operating flow, treasury reserve, custody, settlement, trade execution, or a structural purpose such as holding, escrow or a special purpose vehicle. Each is placed in a jurisdiction chosen for legal certainty, correspondent reach and regulatory fit rather than for proximity or convenience. The whole is designed once, reviewed on a cycle, and changed deliberately.
| Test | Accumulated structure | Designed architecture |
|---|---|---|
| Purpose of each account | Remembered, sometimes | Recorded and current |
| Liquidity position | Assembled on request | Reported |
| Foreign exchange | Happens at payment | Decided in advance |
| Reconciliation | Investigative | Mechanical |
| Concentration risk | Discovered | Chosen and sized |
| Number of relationships | Grows | Usually falls |
The last line surprises people. A designed structure is typically smaller than the one it replaces. That is not an economy measure, it is the point: capability concentrated in relationships that are properly serviced beats capability scattered across relationships that are tolerated.
What the transition costs
Consolidation is unglamorous work and it should be priced honestly before it is started.
Closing a relationship means migrating standing instructions, direct debits, counterparty payment details, payroll routing and any collateral or guarantee arrangements attached to the account. Counterparties must be told, and some will require documentary evidence of the change. Where an account supports an entity in a jurisdiction with local filing requirements, the accountant needs notice. Dormant accounts holding small residual balances are often the most tedious to close, because the institution's offboarding process is designed for the convenience of the institution.
Expect the exercise to run one to two quarters for a mid-sized group, and to absorb more finance attention in the middle of it than in the twelve months before. The return arrives afterwards and it is durable, which is the reverse of the profile that makes projects attractive to approve.
Four tells of a fragmented structure
- Nobody can produce a one-page list of relationships with a stated function for each.
- The group's consolidated cash position is a document that is built rather than a figure that is read.
- At least one account exists because closing it seemed like effort.
- Payment problems are escalated by whoever happens to know somebody at the bank.
Where to begin
Not with a bank. With a list.
Set out every relationship the group holds, the entity that owns it, the currencies it operates in, the average balance, the function it performs today, and the function it was opened to perform. Where the last two columns disagree, the structure has drifted. Where the function column is empty, the relationship is cost without capability.
That single document usually reveals more than a quarter of advisory work, and it is produced internally. Everything that follows, function definition, jurisdictional fit, institutional selection, is a discussion that can only be held properly once the list exists.