Delivering Data Change in a Private Bank: Modernisation Without Losing Trust
The parallel spreadsheet the programme wants to abolish is not obstinacy. It is the relationship manager refusing to place a number in front of a client until they are sure it is right.
Executive Summary
Private banks are being told, as almost every institution is now being told, that they must modernise their data. The instruction is sound in its intent and easy to misread in its framing. It is sound because the manual reconciliations, the spreadsheets passed between desks, and the fragmented client records that many private banks still rely upon are a genuine constraint on service, on control, and on the ability to answer a regulator or a client with confidence. It is easy to misread because the language in which modernisation arrives – platforms, pipelines, single client views, straight-through processing – is borrowed almost entirely from volume and retail banking, where much of the ambition is to remove human judgement from the transaction. In a private bank, human judgement is the service.
This essay argues that data change in a private bank is not a smaller version of the retail programme. It is a different problem with a different centre of gravity. The retail programme optimises a process that millions pass through; the private programme must strengthen the evidence base beneath a few thousand relationships that are, individually, irreplaceable. The tension that must be managed throughout is between what modernisation can improve and what it can quietly destroy: the discretion, the judgement and the relationship-manager trust on which the franchise rests.
The recommendation is not to modernise less. It is to modernise in a particular order, with an explicit register of what is being changed and what is being protected, and with the senior relationship managers treated as the arbiters of whether trust is being preserved rather than as obstacles to be worked around.
Why A Private Bank Is Not A Small Retail Bank
The pattern I have observed across institutions is that data programmes are designed as if all banks were the same bank at different scales. They are not. A retail bank’s relationship with a customer is mediated almost entirely by systems; the customer rarely knows the name of anyone inside the institution. A private bank’s relationship is mediated by a person. The client knows their relationship manager, has often known them for years, and stays or leaves substantially on the strength of that individual bond.
This changes what data is for. In the retail world, data exists to run the process at scale and to remove the need for judgement at the point of service. In the private world, data exists to arm the judgement of the relationship manager and the teams around them. It is an input to a human decision, not a substitute for one. A modernisation programme that does not grasp this distinction will optimise the wrong thing with great efficiency.
Four characteristics define the private bank and none of them survive being treated as a retail edge case.
- Discretion. The institution holds information whose value depends partly on how tightly it is held. Widening access to client data, which is often the explicit goal of a single-view programme, is not a neutral technical act. It is a change to who inside the bank can see what, and it touches the core promise the bank makes.
- Judgement. The service is bespoke. Two clients with identical balance sheets may be advised differently because their circumstances, temperaments and intentions differ. Data that flattens this into a segment misrepresents the very thing the client is paying for.
- Relationship primacy. The relationship manager is the franchise. Institutional knowledge about a family, a business, a history of decisions, sits substantially in their head and their notes, not in a system. A programme that does not find a way to respect and capture that knowledge is removing capability while believing it is adding it.
- Reputation. In a volume bank an error is a defect rate. In a private bank a single wrong figure shown to a client can end a multi-generational relationship. The tolerance for the visible mistake is close to zero, and that fact must shape how change is introduced.
The retail programme optimises a process that millions pass through. The private programme must strengthen the evidence beneath a few thousand relationships that are, individually, irreplaceable. The engineering may look similar; the risk profile is not.
What Modernisation Actually Threatens
It is worth being precise about what is at risk, because the threat is rarely the one named in the programme documents. The stated risks are the familiar ones: migration failure, data quality gaps, delivery slippage. The real risk is subtler and is almost never on the register. It is the erosion of the conditions under which relationship managers trust the numbers in front of them and feel able to sit in front of a client.
Consider the ordinary scenario of relationship-manager management information. Today, in many private banks, the senior relationship manager does not fully trust the central report. They keep their own view – a spreadsheet, a set of notes, a mental model built over years – and they reconcile the official figure against it before they act. This is inefficient and it is exactly what a modernisation programme wants to eliminate. But the private habit of a parallel personal record is not mere obstinacy. It is a control. It is the relationship manager refusing to place a number in front of a client until they have satisfied themselves it is right.
If the programme removes the parallel record without first earning the relationship manager’s confidence in the central one, it has not improved the control environment. It has removed a control and replaced it with a leap of faith. The relationship manager will either quietly rebuild their spreadsheet, in which case nothing has been gained and trust in the programme has been spent, or they will act on figures they do not trust, in which case the bank has increased the probability of the one error it can least afford.
“The parallel spreadsheet the programme wants to abolish is not obstinacy. It is the relationship manager refusing to place a number in front of a client until they are sure it is right.”
The same logic applies to client confidentiality. A single client view that makes a household’s full picture visible to more of the organisation may be operationally sensible and may still be a breach of the implicit promise on which the relationship was formed. Under the data protection regime that now governs this – purpose limitation, data minimisation, the discipline of holding only what is needed for a defined use – the wider question is not merely whether access is lawful but whether it is consistent with what the client believed they were agreeing to. In a private bank those two tests are not the same, and the second is the one that governs reputation.
What To Change
None of this is an argument for standing still. The manual estate of many private banks is a genuine liability, and it is worth naming the things modernisation should unambiguously improve.
- The evidential base beneath advice. When a relationship manager advises, the underlying position, exposures and history should be available, consistent and current, so that judgement is exercised on good information rather than on whatever could be assembled by hand that morning.
- Reconciliation and control. The reconciliations that today consume senior time and introduce operational risk are precisely the work that should move into a controlled, automated, auditable process. This is the strongest part of the modernisation case and should be led with.
- Regulatory and reporting confidence. The ability to answer a regulator, or the bank’s own risk committee, promptly and consistently is a real capability gain, and one that does not touch the client relationship directly. It is low-risk ground on which to demonstrate that the new estate can be trusted.
- Resilience and continuity. Knowledge that sits only in one relationship manager’s head is a concentration risk to the franchise. Capturing enough of it, respectfully, so that a client is not orphaned when a relationship manager retires is a genuine improvement – provided it is done as augmentation, not surveillance.
The common thread is that the safest and most valuable early modernisation targets are the ones that strengthen the relationship manager’s hand without altering the relationship itself. Reconciliation, control and reporting improve the bank’s confidence in its own numbers. That confidence is the foundation everything else is built on.
What To Protect
Against the list of what to change sits a shorter, firmer list of what must not be traded away, whatever the efficiency case.
| Protect | Why it is non-negotiable | What a careless programme does to it |
|---|---|---|
| Discretion over client data | The promise of tight holding is part of the product | Widens access in the name of a single view |
| Relationship-manager judgement | The bespoke service is the franchise | Replaces judgement with a segment or a score |
| The right to a trusted number | A visible error can end a relationship | Removes the parallel control before earning trust |
| Institutional knowledge | It is capability, not inefficiency | Discards notes and context as legacy clutter |
The discipline this table is meant to enforce is simple: every proposed change should be tested not only against the operational benefit it delivers but against the column of things that must be protected. Where a change strengthens the evidence base without widening access or displacing judgement, it is low risk and should proceed. Where a change delivers efficiency by widening access, flattening judgement, or removing a control before its replacement is trusted, it should be slowed down and reconsidered, however attractive the business case looks on a slide.
A Scenario: The Bespoke Mandate
It helps to make this concrete. Consider a long-standing client with a bespoke arrangement: a concentrated holding they will not sell for reasons that are personal rather than financial, a lending facility structured around that holding, and a set of understandings built up over two decades between the family and a senior relationship manager. None of this is unusual in a private bank. All of it is invisible to a data model designed for the general case.
Now imagine the modernisation programme arrives and, quite reasonably, wants to bring this client into the standard client view so that exposures can be monitored centrally and reported consistently. The engineering is straightforward. The risk is entirely in what the standard view cannot hold. The model will record the concentrated position as a risk to be flagged. It will not record that the concentration is deliberate, understood, and the subject of a long-settled conversation. If the central system now generates an alert, or worse, prompts a well-meaning colleague who does not know the history to contact the client about reducing the position, the bank has not improved its control. It has manufactured an intrusion into a relationship it spent twenty years earning.
The lesson is not that such clients should be kept out of the modern estate. It is that the estate must be designed to carry context, not only figures, and that the relationship manager must remain the interpreter of that context. A field for the exposure is easy. A disciplined place for the reason the exposure is intended, visible to the right people and governed with the same care as the number itself, is the harder and more important piece of design. Where the programme cannot yet carry the context, it must not act on the figure alone.
Sequencing: The Real Instrument Of Control
If the argument so far is accepted – that the danger is to trust rather than to the technology – then the primary instrument for managing that danger is sequence. The order in which change arrives determines whether senior relationship managers experience modernisation as something being done for them or to them, and that experience, more than any feature, decides whether the programme succeeds.
The pattern that works is to earn trust before spending it.
- Begin where the client is not touched. Lead with reconciliation, control and regulatory reporting. These deliver visible value, reduce operational risk, and let relationship managers watch the new estate produce numbers that match their own before anything depends on it. Trust in the central figure is built by letting people check it against the parallel record they already keep, not by abolishing the parallel record on day one.
- Let the parallel record retire itself. Do not mandate the removal of the personal spreadsheet. Make the central view good enough, for long enough, that the relationship manager stops reaching for the spreadsheet because they no longer need it. The retirement of the workaround is the evidence that trust has transferred, and it should be read as a milestone, not engineered as a precondition.
- Change access last and deliberately. Any widening of who can see client data should come after the bank has demonstrated it can be trusted with the data it already holds, and it should be justified in the language of client promise, not only operational convenience. Where the data protection regime demands minimisation, treat that discipline as an ally rather than an obstacle: it is the same discipline the client relationship already implies.
- Capture institutional knowledge as augmentation. Approach the senior relationship managers as the source of what is worth keeping, not as the subject of an extraction exercise. Knowledge captured with their authorship is an asset the franchise retains; knowledge captured over their objection is resented and, tellingly, tends to be incomplete.
The reason sequence matters more than any individual design decision is that trust, once spent, is expensive to rebuild, and senior relationship managers are watching the first phase to decide how much of it to extend to the rest. A programme that opens by removing a control it has not yet replaced teaches the most influential people in the bank that modernisation costs them something. A programme that opens by making their reconciliations disappear and their reporting effortless teaches them the opposite, and buys the permission it will need for the harder changes later.
Closing
The temptation in a private bank is to run the programme that worked elsewhere, faster, on the assumption that a private bank is simply a smaller and more complicated retail bank. It is not. It is an institution whose value is concentrated in judgement, discretion and relationships that a system can support but must never presume to replace. Modernisation that respects this is not slower for the sake of caution; it is sequenced so that the bank’s confidence in its own data grows ahead of its dependence on it, and so that the people who carry the franchise experience the change as reinforcement rather than removal.
The measure of success is not the retirement of a legacy platform or the go-live of a single client view. It is that, a year after the change, the senior relationship manager trusts the number on the screen enough to place it in front of a client without checking it against anything else – and that the client, who never saw the programme at all, has noticed only that the service became a little sharper and never once became less discreet.