Public services do not have a data shortage, or an AI shortage. They have a data integration problem. Until good-quality, connected data sits on a shared foundation, AI will continue to remain a slide in a strategy deck, and citizens will keep paying the cost of joining up the state themselves.
Through four key service journeys, this article explores how legislation designed to protect and govern data can inadvertently make it harder to share information, connect services and improve outcomes for citizens.
Ask a room of public sector leaders what is holding transformation back and AI keeps entering the conversation. This is closely followed by skills, ethical assurance and data access. All are real. But most of that conversation is a comfortable distraction.
The reality is AI is a problem about the future. However, the constraint that matters is one that already owned by the estate leaders: inconsistent, duplicated data that is locked inside organisational boundaries.
Any public body that cannot reliably answer ‘what do we already know about this person, household or property?’ is not ready to deploy AI in a citizen-facing service.
The National Audit Office's assessment of the Government Finance Function's Strategy 2030, which guides more than 10,000 government finance specialists, found the function has "established many of the foundations needed to succeed". It also flagged "outdated systems and poor-quality cost data making it harder to take full advantage of artificial intelligence and to track potential savings".¹
In summary, NAO is not saying the government lacks AI tools. It is saying the data underneath is not good enough to use them.
The wider estate tells the same story. Around 28% of the government IT estate is classified as legacy.³ DSIT's State of digital government review found departments use multiple ways of identifying the same people and businesses. This means that records describing the same citizen cannot reliably be matched.⁴ Adding to this, the ONS estimates that the UK public service productivity in 2025 is still 3.1% below its 2019 level. This is despite years of technology investment (the ONS notes these estimates remain under development).⁵
None of this is a technology gap. It is a foundations gap, and the guidance to close it already exists. The Government Data Quality Framework has set out data quality principles and dimensions since 2020. The Technology Code of Practice and GDS Service Standard require services to be interoperable and to use open standards. The problem is not a lack of rules; it is a lack of enforcement and investment in following them.
The most common reason given for not sharing data is "GDPR won't let us". To improve public services, leaders should stop accepting it as an answer.
The statutory framework for lawful public sector data sharing is extensive:
The real barriers are organisational: no named data owner, no agreed definitions, risk-averse information governance with no incentive to say yes, and data sharing agreements that take longer to sign than the service takes to build. A data sharing agreement is not a lawful basis; it documents one. When that happens, the governance model is the problem.
Fragmentation is not an IT problem. It is a service delivery and equity problem, and under the Public Sector Equality Duty (Equality Act 2010, section 149) public bodies must have due regard to its effects. Those most dependent on joined-up services (disabled people, older people, low-income families, children at risk) are the ones most exposed when it fails.
The Department for Transport records 3.07 million Blue Badges held in England at 31 March 2025, with 1.27 million issued during 2024/25.² Under the Disabled Persons (Badges for Motor Vehicles) (England) Regulations 2000, many applicants qualify automatically because of a benefit award that DWP already holds. Yet applicants are routinely asked to upload award letters, re-prove addresses and re-submit clinical evidence that other public bodies already have.
The impact: every one of those uploads is a cost transferred from the state to a disabled person.
The expansion of free school meals to all households on Universal Credit in England from September 2026 makes the gap between entitlement and take-up a live issue. Councils that have auto-enrolled eligible children use data they already hold for housing benefit and council tax support, have shown what is possible.
The impact: Eligible children miss out and schools lose the pupil premium funding attached to them. However, this is the clearest example in public services of a data join that pays for itself.
NHS England publishes data on patients who no longer meet the criteria to reside. The delay reasons are instructive: social care arrangements, housing, rehabilitation, care home availability, transport.⁶ Almost none are clinical. Since the Health and Care Act 2022 placed Integrated Care Systems on a statutory footing and discharge to assess became the default model, the expectation is clear. NHS England now requires every ICS to develop or join a shared care record, and the Adult Social Care Client Level Data return has been mandatory for local authorities since 2023.
The impact: Both are admissions of how much the lack of a single view has been costing.
Almost every serious case review and child safeguarding practice review returns to the same finding: agencies each held part of the picture and nobody held all of it. The national review into the murders of Arthur Labinjo-Hughes and Star Hobson (2022) made the same point, and the Children's Wellbeing and Schools Bill proposes a consistent child identifier to address it. Working Together to Safeguard Children (2023) is explicit that information sharing is essential and that data protection is not a reason to withhold it.
The impact: Without a consistent identifier, a shared foundation to resolve records against it, it will not be sufficient.
What good already looks like: The Tell Us Once and the National Fraud Initiative is a positive example where data sharing across the public sector works. Tell Us Once lets a bereaved family report a death once and have it passed to DWP, HMRC, the Passport Office, DVLA and the local council. The National Fraud Initiative, run by the Cabinet Office under the Local Audit and Accountability Act 2014, matches data across hundreds of organisations to detect fraud and error. Neither is new technology. Both succeed because someone defined the identifiers, agreed the legal gateway and owned the data flow.
Ultimately, joined-up data foundations enable public services to organise themselves around the citizen, rather than expecting the citizen to navigate the boundaries between organisations. When information can be shared securely and responsibly, it stops the onus being placed on individual spending unnecessary time repeating their circumstances and navigating disconnected processes. Instead, it places services in response to citizen needs.
The outcome is a public service experience connected with good-quality, connected data sits on a shared foundation.
Not sure where to start? Get in touch to see how we can help.
---
References