Heritage vs Legacy: When ‘Old’ Means Something Different in Banking Technology
Old and working is heritage; old and constraining is legacy — and the fastest way to lose the room in a centuries-old bank is to call the first one the second.
The Word That Loses the Room
The technology industry has settled on a single word for old systems, and that word is legacy. It is never a compliment. To call a system legacy is to say it is a burden — outdated, brittle, an impediment carried forward from a less enlightened time, something to be escaped. The word arrives pre-loaded with a verdict, and the verdict is always the same: replace it.
In most of the industry that shorthand is harmless enough. But bring it into a bank that has been in business for a century or more, and it does real damage on the first day. Because in an institution of that age, being old is not a defect. It is the brand. The firm’s longevity is its central claim — the reason a client trusts it with wealth meant to outlast them, the reason the name means what it means. When a newly arrived technology leader walks the estate and starts labelling things legacy, the people who built and maintained those systems hear something the leader may not have intended: that the history they are part of is a problem to be cleared away. The room closes. And a technology leader who has lost the room in an institution that runs on continuity has lost far more than a rhetorical skirmish.
In a long-established bank, “old” is not an engineering judgement. It is an identity. Attack it carelessly and you are no longer critiquing a system — you are critiquing the institution’s account of itself.
The error is not that the leader wants to modernise. The estate almost certainly does need to change. The error is conceptual: collapsing two genuinely different things into one dismissive word. Some old systems are heritage. Some old systems are legacy. They are not the same, and the entire credibility of a transformation in a heritage institution rests on being able to tell them apart — and being seen to.
Two Kinds of Old
The distinction is simple to state and easy to get wrong under pressure. A heritage system is old and working. It does its job, it is understood, it is stable, and its age is incidental to its value — or, sometimes, part of it, because decades of use have hardened it against every edge case the business can throw at it. A legacy system is old and constraining. Its age is not incidental; it is the source of the problem. It blocks something the business now needs to do, it concentrates risk in ways that can no longer be justified, it depends on knowledge or skills that are disappearing, and every year it persists the cost of the eventual change rises.
The reason the two get conflated is that on the surface they look identical. Both are old. Both may run on unfashionable technology. Both may be poorly documented and maintained by a small number of people who have been there a long time. An engineer doing a rapid estate review, primed by the industry’s vocabulary, will tar them with the same brush. That is precisely the mistake, and it is expensive in both directions. Label a heritage system as legacy and you propose to spend money and risk stability replacing something that was quietly doing its job — while insulting the people who kept it running. Label a legacy system as heritage and you leave a genuine constraint in place out of misplaced respect, letting risk compound until it forces your hand on its terms rather than yours.
Judging Which Is Which
So the useful question is never “how old is this system?” It is “is this system’s age costing us anything?” A few honest tests separate the two, and none of them is about the technology’s vintage.
- Does it constrain the business? Not “is it modern” but “is there something the firm needs to do that this system prevents, delays, or makes disproportionately expensive?” A system that blocks nothing the business actually wants is not a problem waiting to be solved, however old it looks.
- Where is the risk, honestly? A concentration of operational or continuity risk — a single point of failure, an unrecoverable dependency, knowledge held in one or two heads that are approaching the door — is a legacy signal regardless of how well the thing currently runs. Stability today is not the same as safety tomorrow.
- Is the knowledge sustainable? A system maintained by people whose skills can still be found and renewed is in a different category from one that depends on expertise the market no longer produces. The second is on a clock even if it works perfectly.
- What does waiting cost? For true legacy, delay compounds — the migration gets harder, the risk grows, the options narrow. For heritage, waiting costs little, because the system is not deteriorating in any way that matters. If deferring the change is cheap, it probably is not legacy.
Apply those tests and most estates sort themselves into three groups rather than two. There are the clear legacy systems, where age is actively constraining or risky and the case for change is real. There are the clear heritage systems, working and sustainable, where the right answer is to leave them alone and say so plainly. And there is a middle band — old, working today, but resting on knowledge or dependencies that are quietly eroding. That middle band is where judgement earns its keep, because it is neither a triumphant replacement nor a comfortable status quo; it is a managed transition to be planned before the erosion forces it.
Making the Case Without Insulting the House
Sorting the estate correctly is only half the task. The other half is how you speak about it, and in a heritage institution that is not a matter of tact — it is a matter of whether the transformation survives contact with the organisation.
The move that works is to make the distinction itself public and to use it consistently. When a technology leader visibly separates heritage from legacy — when they say, of a system, “this is old and it works and we are keeping it, and here is why we respect it,” and then say, of another, “this one is old and it is now holding us back, and here is the specific thing it blocks” — they change what the organisation hears. The message is no longer “your past is a problem.” It is “I can tell the difference between what your history got right and what has quietly turned into a liability, and I am only asking to change the second.” That is a message a proud institution can accept, because it honours the thing the institution is proud of while still telling the truth about what must move.
“Respect the systems that earned their keep, name the ones that no longer do, and never let the organisation mistake the second act for contempt of the first.”
The opposite move — the blanket legacy label, the modernisation programme pitched as an escape from the firm’s own past — fails even when the underlying engineering judgement is sound, because it asks the institution to agree that its history is an embarrassment. It will not, and it should not. The people who maintained these systems for decades are not obstacles to the future; they are the reason there is an institution with a future to argue about. A case for change that treats them as dead weight forfeits their knowledge exactly when it is most needed, because the person who knows why the old system does the strange thing it does on the last working day of the quarter is the person you have just insulted.
The Practitioner’s Position
The pattern that recurs across long-established institutions is that the successful modernisers are not the boldest ones. They are the ones who are precise. They refuse the industry’s lazy vocabulary, they do the work of sorting old-and-working from old-and-constraining, and they make that sorting visible so that everyone can see the transformation is discriminating rather than indiscriminate. In doing so they earn the right to be believed when they do point at something and say it has to change — because they have already demonstrated, repeatedly, that they will also point at something and say it should stay.
Old is not a verdict. In a bank whose whole proposition is that it has endured, treating age as automatic evidence of failure is not just poor manners; it is poor engineering, because it discards information the estate is trying to give you. Heritage is what to protect. Legacy is what to change. The discipline of a technology leader in these institutions is to hold the two apart under pressure — and to make the case for the second without ever letting it sound like contempt for the first.