Why not just archive the whole database and switch the system off?

Because a database dump is not evidence. It preserves the rows without the context that made them a record, and it usually cannot be searched or produced in a form anyone would accept later. What survives has to keep its meaning, its retention rule and its access rules, which is a records exercise rather than a backup, in the sense the records management standard ISO 15489 uses.

How much does keeping an old system running actually cost?

More than the licence line suggests. It carries the infrastructure it sits on, the specialists who still know it, the security exceptions it needs to survive audit, and the integration links nobody dares remove. In most estates the licence is the smallest of those four.

What happens to the data that has no reason to survive?

It is deleted, deliberately and on a written rule, which is the part of the exercise that produces the saving and the part organisations postpone. Content kept because nobody was sure is content you pay to store, migrate, search and eventually produce in a dispute.

← Integration and migration

Decommissioning

Retiring an application you still have to keep the records from, and why the shutdown is where most of the cost take-out actually lands.

What decommissioning actually involves

Switching an application off is the last step and the easy one. The work before it is deciding what has to survive the shutdown, extracting it in a form that will still be readable and provable years later, and proving that nothing else was lost.

Done in that order it retires a system permanently. Done in the other order, the old system stays powered on because nobody is willing to be the person who switched it off.

The shutdown is not the project

An application reaches the end of its life long before it is switched off. Somewhere behind it are records that outlive it: contracts still in force, personnel files inside their retention period, invoices supporting postings that an auditor can still ask about.

So the sequence that works is: classify what is inside, decide what has to survive and under what rule, extract it into something that can hold it with its meaning and its retention intact, prove the extraction is complete, and only then switch off. The final step takes an afternoon. Everything before it is the work, and skipping it is why so many estates carry systems that were decommissioned on paper years ago (International Organization for Standardization, 2016).

Where the money is

Decommissioning is the most direct cost take-out available in most estates, because the saving is a line somebody is already paying.

The licence stops. The infrastructure underneath it stops. The specialists who maintained it move to work that matters. The security exceptions it required disappear from the next audit. And the integration links into it, each of which was a small maintenance liability, stop needing to be tested every time something upstream changes.

That is also why it is the first place we look when a client asks where cost comes out. Full detail on the wider case is on the cost take-out page.

What we do

The inventory first: what the system holds, how much of it is redundant, obsolete or trivial, and what carries an obligation. Then the extraction and its proof, then the shutdown, then the removal of everything that existed only to keep it alive.

It runs as a slice rather than a programme, which means the first system proves the method and the second one costs less. The sequencing logic is the same one behind integration and migration generally: nothing irreversible happens on a single weekend.

Most estates have at least one system kept alive purely because of what might be inside it. That one is usually the cheapest place to start, and the saving is visible in the same financial year.

Book a conversation

Decommissioning, asked plainly

Why not just archive the whole database and switch the system off?
Because a database dump is not evidence. It preserves the rows without the context that made them a record, and it usually cannot be searched or produced in a form anyone would accept later. What survives has to keep its meaning, its retention rule and its access rules, which is a records exercise rather than a backup, in the sense the records management standard ISO 15489 uses.
How much does keeping an old system running actually cost?
More than the licence line suggests. It carries the infrastructure it sits on, the specialists who still know it, the security exceptions it needs to survive audit, and the integration links nobody dares remove. In most estates the licence is the smallest of those four.
What happens to the data that has no reason to survive?
It is deleted, deliberately and on a written rule, which is the part of the exercise that produces the saving and the part organisations postpone. Content kept because nobody was sure is content you pay to store, migrate, search and eventually produce in a dispute.