The full cost of an application option includes operating it, making the transition, maintaining any period of coexistence, and eventually leaving the resulting arrangement. Retention and replacement need comparison over the same scope and time horizon so that a saving in one category does not conceal a cost elsewhere.
Business as usual supplies a benchmark for that comparison. It describes the consequences of continuing current arrangements, including known operating burdens and required remedial work. Keeping it visible does not make it an acceptable final choice when it fails the objectives.
Start with an observed baseline
License purchases show what was bought. Usage evidence shows what is being consumed. Applicable commercial terms determine which entitlements can be changed or transferred. These records answer different questions and need reconciliation.
GAO’s 2024 software-license work calls for comparing licenses in use with purchases to identify optimization opportunities. It does not establish a saving percentage or an entitlement for another organization.
Hosting, support effort, batch operations, external services, and recovery also belong in the baseline. Shared costs need a stated allocation method so that removing one application is not assumed to remove an expense that other applications still require.
Compare credible scopes
Retaining and governing, stabilizing, upgrading, rehosting, replacing, outsourcing, and retiring can have different eligibility conditions. An option that cannot preserve an essential capability should explain that exclusion instead of receiving an artificially attractive cost estimate.
The common comparison should identify the capability being delivered and the period being considered. A replacement estimate covering implementation alone cannot be compared fairly with several years of current operating expenditure.
HM Treasury’s 2026 Green Book uses business as usual as an appraisal benchmark. That public-sector context supports the comparison principle here; its formal rules are not imposed on every private application decision.
Transition creates its own costs
Discovery, conversion, validation, integration, training, data correction, and cutover are distinct activities. Some occur before the replacement delivers any operating benefit. Others continue while the old and new arrangements coexist.
Coexistence can include duplicate licenses, retained infrastructure, two support processes, reconciliation work, and delayed retirement. Its duration should be an explicit assumption with an owner responsible for resolving the dependencies that keep it open.
Take a constructed example in which an application’s hosting allocation falls after migration, but the shared server remains necessary for another application. The allocation has changed; the organization’s cash expense might remain. The saving mechanism must identify the cost that actually disappears.
Realized outcomes need a measurement period
The NAO’s May 2016 shared-service-centres review reported £90 million in customer savings against £94 million in costs over the first two and a half years. Most planned outsourcing transfers had occurred, while only two of 26 planned customers had moved to a single operating platform.
Those are historical findings about a particular programme at its review date. They show why service transfer, platform transformation, costs, and savings need separate accounting. They do not establish that all outsourcing has the same result.
Forecast benefits should remain forecasts until observed. A report should state which period and population its realized figures cover.
Make uncertainty capable of changing the decision
Large assumptions deserve ranges and verification: transition duration, retained-system life, usage growth, supplier terms, and the effort needed to preserve exceptions. Unmonetized effects also need explicit treatment rather than disappearing because a spreadsheet lacks a price.
The preferred option should identify which assumptions would reverse its advantage and which action could resolve them. A cost comparison is useful when it explains how the expected outcome will occur and what evidence would require the decision to change.