The Modernisation Programme That Never Ends
The result is not modernisation. It is accretion.
The Programme That Outlives Its Sponsors
Every large organisation I have worked alongside has one: the modernisation programme that has been “eighteen months from completion” for the better part of a decade. It has outlived two chief information officers, three names, and at least one reorganisation. It is always about to finish. It never does.
We tend to explain this in the language of execution — poor estimation, scope creep, underfunded testing, the usual charge sheet. Those failings are real. But they are symptoms. The deeper reason legacy modernisation never ends is that we insist on treating a permanent condition as a temporary project. The programme structure itself — with its start, its end, its benefits case that closes — is the disease it claims to cure.
Legacy Is Not a Backlog. It Is a Tense.
The word legacy misleads because it sounds like a category of system — the old ones, over there, that we will one day be rid of. It is more honest to treat legacy as a tense. Every system becomes legacy the moment it goes live and the organisation moves on around it. The bright new platform we are standing up to retire the mainframe will itself be somebody’s modernisation headache before it is fully paid off.
“There is no state called “modernised”. There is only the rate at which an organisation renews what it runs.”
Once you accept that, the framing of the never-ending programme falls apart. A programme is the wrong instrument because a programme is defined by its completion; it exists in order to be closed. But keeping a large technology estate current is, by its nature, work that does not close. You no more finish it than a city finishes maintaining its roads.
Why the Great Replacement Keeps Failing
The dominant model is still the great replacement: stand up a multi-year programme, freeze the target, migrate everything, decommission the old, cut over, celebrate. I have watched several of these at close range, and the pattern of failure is remarkably consistent.
- The target is frozen at the start, when the organisation knows least, and then defended for years after the business has moved on. By cutover the programme delivers what was needed at kick-off, not what is needed now.
- The value sits entirely at the end. Nothing is realised until the final cutover, so every delay compounds and every pound spent stays at risk until the last moment — the worst possible risk profile.
- The old system cannot actually be switched off, because a handful of edge cases, regulatory reports or forgotten integrations still depend on it. So the estate grows: now you run the old and the new.
- The people who understood the original system retire or move on during the years the programme runs, taking with them the undocumented knowledge the migration needed most.
The result is not modernisation. It is accretion. We rarely remove; we mostly add. Each great replacement leaves behind a sediment of half-retired systems that the next programme will be chartered to clear away — and so the cycle renews itself, always in the name of ending it.
A Short Anatomy of Accretion
It is worth being precise about how the estate grows even as we spend to shrink it, because the mechanism is not stupidity; it is a chain of individually reasonable decisions.
A migration reaches ninety per cent. The final tenth is the awkward tail — the bespoke integration nobody fully documented, the quarterly regulatory extract that only runs off the old database, the downstream team that will not be ready until next year. Switching off the old system would strand them. So the sensible, defensible call is made: keep it running a little longer. That call is correct every time it is made, and its cumulative effect is an estate that never gets smaller. The old system settles into a permanent semi-retirement — still consuming licences, still needing patching, still holding one thread of the business hostage. Multiply that across a decade of programmes and you have the true shape of most enterprise estates: not layers of the new replacing the old, but strata, each one preserved by a reasonable decision not to quite finish the job.
The Honest Reframe: Renewal, Not Replacement
The organisations I have seen make genuine progress did something quietly radical: they stopped trying to finish. They replaced the programme with a capability. The distinction sounds semantic; in practice it changes everything.
A replacement asks, “How do we get from the old estate to the new one?” A renewal capability asks, “What is our sustained rate of change, and is it faster than the rate at which our estate decays?” The first question has an end. The second is a permanent operating discipline — and it is the right one.
Concretely, the shift looks like this. Rather than freezing a distant target, you carve the estate into pieces that can be renewed independently, strangling the old system component by component while it keeps running — replacing the engine in flight rather than building a second aircraft and hoping to leap across. Value arrives as a steady stream instead of a single terminal event. Risk is spread across many small, reversible cutovers rather than concentrated in one that is not. And the funding stops being a fixed capital programme with a closing date and becomes a standing allocation — the cost of staying current, treated the way we treat the electricity bill rather than a one-off building project.
Stop asking when the modernisation will be finished. Start asking whether your rate of renewal exceeds your rate of decay. Only one of those questions has a useful answer.
The Objection, and the Answer
The obvious challenge from the finance director is that a programme with no end date is a blank cheque. It is fair, and it deserves a straight answer.
The answer is that the never-ending programme is already the blank cheque — we have simply disguised it as a series of “final” pushes, each one signed off on the promise that this time it will close. Continuous renewal is not less disciplined than the great replacement; it is more. It forces you to justify value every quarter rather than once, at the outset, on a business case nobody reopens. It makes the true running cost of the estate visible instead of burying it inside a capital programme that is quietly re-chartered every few years. The blank cheque was never the honesty of “this never ends”. It was the fiction that it was just about to.
What This Asks of Leaders
None of this is, at root, a technical argument — which is why it so rarely gets made in the room where it matters. Continuous architectures, incremental migration, component-by-component replacement: the engineering community has understood these approaches for years. The blockage is governance and psychology.
Leaders like programmes because programmes end, and things that end can be celebrated, put on a slide, attached to a bonus. A capability that simply persists offers no ribbon to cut. To fund renewal properly, an executive has to give up the satisfaction of declaring victory and accept instead the quieter discipline of a system that is never done but never allowed to rot. That is a harder story to tell a board. It is also the only one that has ever been true.
The programme that never ends is not, in the end, a failure of execution. It is what happens when we organise — often with great competence — around the wrong idea: that the work will one day be over. The organisations that escape it are not the ones that finally finish. They are the ones that stop expecting to.