Vantage Department: a 20% cut across 118 services
Vantage is a government department facing a 20% reduction in its IT run budget from April. Its service catalogue lists 118 services. There is no unit cost data: costs are held by team, not by service, and the finance business partner estimates it would take six weeks to build service-level costs. Twenty-two services have fewer than 15 active users, and at least three of those are statutory - the department is legally obliged to provide them, though usage is low. The minister has said there must be "no visible reduction in service quality". Nine months remain before the cut takes effect, and one union has already raised concerns about the effect on front-line staff.
1. How would you decide what to retain, change and retire? Set out the decision logic, not just the data you would collect.
Read the worked answer
The minister's constraint is unhelpful but not meaningless - it should be read as "no reduction in quality for services that matter to citizens and staff", which forces an explicit definition of what matters rather than a blanket protection.
Decision logic, applied to each service:
- Obligation filter first. Statutory or contractual services are non-discretionary regardless of user count. Low usage plus statutory obligation usually signals a demand-generation or accessibility problem, not a candidate for retirement - cutting them creates legal risk that dwarfs the saving.
- Then cost-to-serve against value. Six weeks to build service-level cost is worth spending, because without it every conversation is anecdotal and the loudest directorate wins. Use activity-based allocation, not perfect costing.
- Segment the remainder into: retire (redundant, superseded, or nobody can name the owner); consolidate (duplicate services across teams); re-price or re-channel (move low-volume, high-touch requests to self-service); retain and protect.
- Test each candidate against risk: what breaks, for whom, and what is the recovery path if we are wrong.
Sequencing: pursue the retire and consolidate segments first, because they fund the protected services and buy credibility for harder decisions later. A 20% target met entirely by visible service withdrawal is a political failure even if it is arithmetically correct.
2. How would you plan and communicate this so the decision survives contact with stakeholders?
Read the worked answer
Two failures to avoid: announcing a list, and running an open consultation with no criteria. Both end with the decision made by whoever shouts loudest.
The plan:
- Separate policy from procedure. The policy decision - the 20% target and the obligation-first rule - belongs with the minister and the permanent secretary. The selection of individual services is a procedure that should be delegated to a panel with published criteria. Mixing them means every service dispute escalates to a politician.
- Publish the criteria before the list, and state who decides and how appeals are heard. Predictability is what makes a cut survivable.
- Engage the union early and on the real issue - which is usually roles and workload, not service lists. Front-line staff will know which low-usage services are actually essential and which are dead; that intelligence is cheap and it also gives them a stake in the outcome.
- Give affected service owners a named contact and a closure plan covering data retention, user migration and support handoff.
- Use the continual improvement model explicitly: where are we now (cost and demand data), where do we want to be (the target and protected set), how do we get there (initiatives and dates), did we get there (review at three and nine months). A visible iteration point makes it easier to correct mistakes without losing the programme.
Keep one message consistent across all audiences: the target, the criteria, the protected set, and the appeal route.
3. What would you measure, and what unintended consequences should you watch for?
Read the worked answer
Measure the outcome the cut is supposed to deliver and the damage it might cause, side by side:
- Run cost per service and per user - the core number, and the reason to build the cost model.
- Demand and channel mix for retained services, to catch suppressed demand being pushed to the contact centre or to informal local workarounds.
- Service availability and time to restore for the protected set, since cuts often show up as fragility rather than as visible withdrawal.
- User effort and complaint themes for migrated or self-serviced services.
- Staff workload and overtime - the union's genuine concern, and an early warning that the saving has been transferred rather than removed.
Unintended consequences to watch: savings appearing in IT while reappearing as shadow IT spend in directorates; low-usage statutory services quietly degrading until they fail an access test; key-person dependency rising as teams shrink, so the next absence becomes an outage; and cost data being optimised by reallocating shared costs rather than by changing anything real. Review at three months with authority to reverse specific decisions - a plan with no reversal mechanism will hide its own errors until they are expensive.