A large estate has twenty, fifty, sometimes hundreds of systems of varying age. It cannot be modernised at once, and the useful question is not which is oldest but which is costing the most.
Age is a symptom. Plenty of old systems are cheap, stable and entirely adequate, and modernising one of those is money spent to arrive where you already were.
Five criteria
1. Cost to keep running
Licences, hardware or hosting, and the specialist support contract that exists because nobody internal can maintain it. This is the number that already appears in a budget and therefore the easiest to establish.
Include the hidden line: how much engineering time goes into keeping it alive versus into anything that moves the business forward.
2. Security and supportability exposure
Is the platform receiving patches? Is the runtime supported? Can you still buy the hardware, and could you rebuild it if the machine died tomorrow?
A system with no security updates is not a modernisation candidate; it is an incident waiting for a date. This criterion tends to override the others when it is genuinely triggered.
3. Regulatory exposure
Does it hold personal data it cannot adequately protect, produce a report a regulator relies on, or lack the audit trail a rule now requires? DPDP obligations around correction and deletion are a common trigger, because systems built before those rights existed frequently cannot honour them.
An audit finding is the most reliable funding mechanism in enterprise IT, and it should be used honestly rather than manufactured.
4. Change velocity
How much does this system slow down things the business actually wants? Measured properly: the change requests that were declined or deferred because touching it was too risky, and the initiatives that were shaped around it rather than served by it.
This is the criterion most often felt and least often measured. Ask the product or operations side what they gave up.
5. Strategic importance
Is this where you compete, or is it plumbing? An unremarkable general ledger and a bespoke pricing engine deserve different answers even at identical age and cost.
Sequencing
Score each system on pain (the first four) and importance (the fifth), then:
| High importance | Low importance | |
|---|---|---|
| High pain | Start here. This is the case that funds itself. | Replace or retire — do not rebuild plumbing. |
| Low pain | Encapsulate, so nothing else depends on its internals. Revisit later. | Leave alone. Genuinely. |
The bottom-right quadrant is the one organisations struggle with. A stable, cheap, unimportant old system is not a problem to be solved. Modernising it because it is old is the clearest way to spend a budget and produce nothing.
Retire before you replace
On every estate assessment we run, something turns out to support a process that no longer exists — a report nobody reads, an integration to a partner relationship that ended, a module for a product line discontinued years ago.
Ask what each system is for before asking what should replace it. Decommissioning is the cheapest modernisation available and it is almost never on the initial list.
Encapsulation as the default first move
For most systems in a large estate, the highest-value first step is not to change them at all. It is to put a documented interface in front of them and point everything else at that.
Nothing inside changes. What changes is that the rest of the estate stops depending on internals, integrations stop being bespoke each time, and any future decision about that system becomes independent of every other system. It costs a fraction of any structural change and it converts a tangled estate into a set of separable problems.
Making the case
Boards fund consequences, not vintages. "The platform is unsupported" is a fact; "we cannot honour a deletion request, which is now a legal obligation" is a case.
Translate each criterion into its consequence: the annual cost of the workarounds, the change the business wanted and did not get, the audit finding, the specific exposure if the one person who understands the system retires. That last one is often the most persuasive and the least comfortable to say out loud.