Shared services: which contract costs will actually stop, and when?

Shared services promise a simple idea. Departments stop running their own back-office systems and move onto a common platform. Fewer systems, fewer contracts, lower costs. Achieving that takes more than choosing the platform. The savings only arrive when the old costs actually stop.

In March 2026 the National Audit Office (NAO) published Update on government shared services, a value for money report on the government’s Shared Services Strategy. The NAO accepts that the strategy offers “potential for significant efficiencies”. But it also concludes that unresolved gaps in ownership, technical leadership and the governance of interdependencies mean that, unless they are addressed urgently, government will not achieve the full benefits it set out to achieve and may incur increased costs.

Most of the NAO’s findings are about central governance, which individual organisations can’t fix alone. The commercial side is different. Whether you are a department, an arm’s-length body or a wider public sector organisation considering a shared platform, one question sits within your control: which of your existing costs will actually stop, and when?

What the NAO report says

The report is worth reading in full. These findings are the most relevant to anyone building or testing a shared services savings case:

• The ambition is large. In March 2021 Cabinet Office refreshed its delivery model, aiming for five cloud-based shared service centres serving five clusters of departments, operational by 2028. The NAO notes that at the time Cabinet Office saw a window of opportunity because many legacy systems were nearing the end of their service contracts.
• Nobody holds all the levers. The NAO found there is “no single owner” in the centre of government with a clear mandate to secure departmental onboarding, which leaves buy-in uncertain and timelines at risk.
• Weak technical leadership has a cost. The absence of a strong technical lead has resulted in inconsistent ERP configurations across clusters. Bringing data together will need further processes and interfaces to be built.
• Interdependencies are poorly managed. At least 25 other digital change programmes across government interact with shared services. The NAO describes the management of these interdependencies as inadequate.
• Benefits can be double counted. Clusters raised concerns that other change programmes had included the same benefits already registered by shared services in their own plans.
• Changing course is expensive. When the central Applicant Tracking System programme was reset in October 2025, Cabinet Office estimated the cost of the reset at £26 million to £38 million, including contract exit costs and continuing provision of the current service.
• Old systems should be switched off. One of the NAO’s recommendations is that clusters identify and discontinue other systems they are still spending money on where the new ERP could do the job instead.

That last point is the commercial heart of the issue. A new platform adds cost from day one. The savings depend on removing what it replaces.

Where the savings case usually leaks

Business cases for shared services tend to be built on gross savings: the full cost of the systems being replaced. In practice, contracts don’t stop neatly on the go-live date. The common gaps are:

• Overlapping licences. Users are licensed on the new platform while licences on the old system run to the end of their term, often because nobody aligned the two.
• Extended support agreements. Legacy systems kept alive past their planned end date usually need extended or bespoke support, which is rarely cheap and rarely competed.
• Services that outlive the requirement. Hosting, integration, managed service and data services stay contracted after the need has moved to the new platform.
• Notice periods and minimum terms. A contract that needs twelve months’ notice doesn’t stop when the migration finishes. It stops twelve months after someone serves notice.
• Exit and transition charges. Data extraction, handover support and early termination fees can be significant. The NAO’s own example shows that exit and continuity costs are real line items, not rounding errors.

Warning signs in your own position

• Nobody holds a single register of the contracts the new platform is meant to replace.
• The savings case assumes legacy costs stop on the go-live date.
• A legacy contract has been renewed or extended “to cover the transition” without a defined end date.
• Onboarding dates have moved, but renewal and notice dates haven’t been reviewed.
• The same savings appear in more than one programme’s business case.

If any of these apply, the savings in the business case are probably overstated, or at least arriving later than planned.

Commercial due diligence before you move

This work doesn’t need to wait for a final platform decision. It is most useful early, while there is still time to shape renewals and notice. A practical approach has six steps:

1. Build the contract register. List every contract, licence and support agreement linked to the functions moving to shared services, including smaller agreements held outside central IT.
2. Record the key dates and terms. For each one, capture expiry, notice period, break clauses, auto-renewal terms, minimum commitments and any termination charges.
3. Map each cost to a migration milestone. Link every contract to the point in the onboarding plan when it is no longer needed, and to the point when notice must be served to stop it.
4. Plan co-terming and short extensions. Where a renewal falls before migration, negotiate a short, flexible extension or co-term with the migration date instead of committing to a full new term.
5. Estimate parallel running honestly. Some overlap is necessary for testing, data migration and cutover. Size it, time-limit it and put it in the business case.
6. Check future purchasing requirements. Before buying anything new, confirm whether the shared platform will provide it, so you don’t add fresh commitments that later become another overlap.

A simple cost stop timeline

The output of this work can be a single table that answers the question directly. For each contract, it shows when the cost could stop and what stands in the way:

This is a template, not a model of any real organisation. The rows are examples; your register will have its own list.

The key column is the earliest realistic stop date. It is the later of the migration date plus any necessary parallel running, and the first date the contract can actually be ended. Where those two dates are far apart, there is either a negotiation to have or a saving that will arrive later than the business case assumes.

Gross savings, net savings and timing

A credible savings case separates three things:

• Gross saving: the annual cost of the contracts that will eventually stop.
• Net saving: the gross saving less necessary parallel running, exit and transition charges, and any new costs the move creates.
• Timing: the year in which each saving is actually released, based on the cost stop timeline rather than the go-live date.

Reducing avoidable overlap can release real budget. But the case only holds up if it accounts for the costs that are genuinely unavoidable. A smaller, well-evidenced net saving is more useful to a finance director than a large gross figure that slips a year every time an onboarding date moves. And as the NAO’s findings on double counting show, benefits should be recorded once, in one place.

Frequently asked questions

What did the NAO conclude about government shared services?
In its March 2026 report, the NAO said the Shared Services Strategy offers potential for significant efficiencies. It also found that gaps in ownership, technical leadership and the governance of interdependencies mean government may not achieve the full benefits and may incur increased costs unless they are addressed urgently.

When should we start reviewing contracts for a shared services move?
As early as possible, and ideally before any legacy renewal falls due. Notice periods and renewal dates often set the earliest point at which a cost can stop, so they need to be known before migration dates are fixed.

Is parallel running a sign of poor planning?
No. Some overlap is necessary to test, migrate data and cut over safely. The aim is to plan it, time-limit it and include it in the business case, rather than let it run on by default.

What is the difference between gross and net savings?
Gross savings are the costs that will eventually stop. Net savings deduct parallel running, exit and transition charges and any new costs. Net savings, profiled by year, are the figure to rely on.