The System Nobody Can Replace: Why Legacy Outlives Every Business Case
Everyone agrees the old system has to go, and every few years the case is written again and quietly shelved. That is not indecision. The business case for replacement is genuinely weak — and the cost of keeping it never appears where anyone is looking.
Most organisations have one. A system that has been there longer than almost everybody who uses it, which everyone agrees should be replaced, and which is still running.
Every few years the case is written again. It is well argued and expensive — and it goes quietly back in the drawer.
1. The Case for Replacement Is Genuinely Weak
The benefits are avoided cost and reduced risk. Both are hypothetical, both arrive years later, and neither shows up as revenue.
Against that sits a system doing its job today. It processes the transactions, it closes the month, it has not fallen over.
A board choosing growth over replacement is not being short-sighted. On the numbers in front of them, they are being sensible.
2. The Risk Is Not Symmetrical
Leaving it costs money slowly and predictably. Replacing it can stop the organisation trading for a fortnight.
Executives price that difference correctly, even when they cannot articulate it. A slow, familiar cost is easier to carry than a fast, unfamiliar one — particularly when the fast one has your name against it.
Which is why these decisions rarely fail on analysis. They fail on who is willing to sign.
3. Nobody Knows What It Actually Does
Underneath sits the real problem. Twenty years of decisions are encoded in that system — pricing rules, exceptions, regional variations, fixes applied during a crisis nobody now remembers.
Very little of it is written down anywhere else. The system is not an implementation of the specification. It has become the specification.
Every year that passes, a few more of the people who remember why retire, and the cost of understanding it goes up.
This is also why replacement estimates are unreliable. You are not costing a build — you are costing an archaeology exercise, and nobody can size that honestly until it is under way.
4. The Cost Shows Up Somewhere Else
Look for it in the IT budget and it seems tolerable — an ageing platform, a maintenance contract, a small team who know it well.
The rest is distributed. It is in the manual steps around it, the reconciliations, the exceptions handled by people, the six weeks it takes to change something a competitor changes in a day.
The price of a legacy system is not what it costs to run. It is what it stops you deciding.
5. Replacement Is Not the Only Shape
Most cases fail because they propose the largest possible version of the change. One programme, one date, everything at once.
There is usually a less dramatic route. Put an interface in front of the old system, move one capability out at a time, and let the original shrink until switching it off is unremarkable rather than terrifying.
Start with the capability that has the cleanest boundary. Usually that is something at the edge — reporting, or a customer-facing view — rather than the ledger sitting in the middle of everything.
It takes longer and makes for a poor announcement. It is also considerably more likely to finish.
6. Sometimes Keeping It Is the Right Answer
This is the part that gets skipped. For a stable process, in a business that is not changing, an old system that works may be exactly the right thing to keep.
The problem is not organisations that decide to keep them. It is organisations that never decide at all, and discover in year fifteen that the choice has been made for them.
7. If You Are Keeping It, Buy Time Deliberately
A decision to wait is only defensible if it comes with work attached:
- write down the business rules while the people who know them are still here
- stop adding new functionality to something you intend to retire
- get the data out into a form you can move and read without it
- remove the single points of knowledge, one person at a time
None of that is glamorous. All of it lowers the cost of the decision you will eventually have to take.
8. Say Which One You Are Doing
The worst outcome is neither. A half-funded modernisation leaves two systems and twice the interfaces — and nobody willing to switch the first one off.
Whatever gets chosen, the organisation should be able to say plainly which it is — replacing, strangling or keeping.
Most organisations never choose to keep their legacy systems. They simply keep not choosing to replace them, one year at a time.