Decommissioning

The most commonly abandoned step in enterprise IT. The new system launches, attention moves on, the old system continues running because switching it off is nobody's priority and carries risk with no visible reward — and the saving that justified the programme never materialises.

The work involved, which is why it gets deferred: confirming nothing still depends on it, which requires monitoring for actual usage rather than asking; extracting and preserving data that must be retained for regulatory periods, in a form that remains readable and queryable without the application; providing access for the residual cases — historical enquiries, audits, disputes — which is usually solved with an archive and a simple search interface rather than by keeping the system alive; revoking integrations and credentials; and reclaiming licences, which is where a large part of the saving actually sits.

The technique that de-risks it and is chronically underused: switch it off temporarily first. Take it offline for a defined window, monitor for complaints and errors, and restore if something surfaces. That converts a feared irreversible act into a reversible experiment, and it finds the forgotten dependency far more reliably than an inventory review.

The governance point: make decommissioning a funded, scheduled deliverable of the modernisation programme with the same status as go-live, and report retirements rather than migrations. Otherwise the estate only ever grows.