Summary

  • On 29 October, the UK government clarified that legacy can include an organisation’s lack of capability or capacity to maintain or improve a digital service, not only ageing or unsupported technology.
  • The Cloud Challenge Book treats SaaS, existing licensed products and cloud services as possible remedies, while saying services already in cloud can still need an upgrade.
  • The 2025 State of Digital Government Review found that most public bodies were still building comprehensive asset registers; about 15% of survey respondents could not estimate the size of their legacy estate.

A cloud address is no longer an adequate answer to a legacy question. The UK government’s first Cloud Challenge Book already said some legacy systems run in cloud. The clarification added on 29 October goes further: a digital service can count as legacy when the public organisation responsible for it lacks the capability or capacity to maintain or improve it.

That shifts the argument away from the age of a server and toward the organisation that has to keep the service working. The government says upskilling existing staff can help, but cannot meet the challenge at the scale required. Its four legacy workstreams call for an automated view of assets, including outsourced services; faster treatment for lower-risk services; better ways to modernise complex and high-risk systems; and upgrades to services already running in cloud. It also wants new services built in ways that do not create the next layer of legacy.

The menu of remedies is deliberately broad. The clarification names software-as-a-service, existing licensed products and cloud services as relevant. Where a remedy depends on cloud infrastructure, the government says its future procurement focus is infrastructure-as-a-service, platform-as-a-service and enabling services. A contract can supply software or operations. It cannot make the public owner’s responsibility disappear: someone still has to understand what the service does, decide what must change, inspect what an outsourced provider is doing and carry the service through a change of supplier.

The inventory problem predates this new definition. The January 2025 State of Digital Government Review said most public-sector organisations were at an early stage of building comprehensive digital-asset registers. Available legacy data was often estimated rather than exhaustive, and about 15% of survey respondents could not estimate the size of their legacy estate. Its estimate of around 28% applied to central-government organisations’ technology estates and ranged from 10% to 60%. The Cloud Challenge Book uses the phrase “28% of our estate” without publishing a matching scope or method.

Those figures cannot be treated as a like-for-like update.

That gap matters because a capability-based definition is harder to close than a server list. A department may replace a supported product and still lack staff, documentation or authority to improve the service. A cloud supplier may take on operating work, but procurement still has to say which decisions the public body retains, what evidence the supplier must provide and how the relationship changes when the service needs to move.

The next commercial route is also unfinished. The government says the National Cloud Infrastructure Programme procurement is expected in early 2027, subject to design and approvals. Its five-to-ten-year horizon is a planning assumption for supplier feedback; the duration, structure and nature of future commitments are undecided. That leaves time to make “legacy” an operating test rather than a migration label.

A useful register would join each service to its public owner, technology, supplier, maintenance capability, known gaps and funded next action. Without that link, a migration can look complete while the service remains hard to change.

Official sources