Policing now has the mandate it lacked. January's white paper commits to integrating information held across local and national systems through a National Data Integration and Exploitation Service, to national data standards removing unnecessary barriers to sharing.1 The Problem Book asks industry thirteen questions. Two of them are really one question, and that mandate now depends on it:
Both are usually treated as system problems. They are not. Almost every force has stood up digital evidence and records platforms over the past eight years. What is missing sits between them: a governed layer where data can be joined, interpreted consistently and acted on, without waiting for structural reform to complete.
The National Audit Office concluded in November 2025 that a lack of standardisation across forces and patchy adoption of digital technology have hampered productivity improvement. Similarly, the Policing Productivity Review identified opportunities worth around 38 million hours of police time, with 443,000 officer hours alone tied up in crime recording that duplicated information already held. HMICFRS has been blunter: the former Chief Inspector described the 43-force model as producing inefficiencies, duplication and inconsistencies.
Translated into a shift, an officer investigating a suspect active across three force boundaries submits requests, waits, then reconciles extracts that define the same offence differently. A safeguarding decision relies on partial information because the fuller picture sits in a neighbouring force, or in probation, and reaching it would take days the risk assessment does not have. This is not a failure of professionalism but of information flow, converting warranted demand into avoidable workload.
Cross-force sharing today means an extract, a transfer, a copy. The moment the copy lands, the originating force loses control of it, it begins drifting out of date, and no one can say with confidence who holds what.
The alternative is to stop moving data, granting governed, revocable access instead. Through Lumilinks' utilisation of the Snowflake environment, a force publishes a permissioned view, a defined slice rather than a database, that a neighbouring force, a regional unit or the CPS queries live. Nothing is exported. Access is logged, scoped and withdrawn at will. Where two organisations need only establish whether an individual appears in both datasets, that match returns without either side exposing its records.
Standardisation is the harder half, and where the white paper's ambition will be won or lost. Offence, risk flag and vulnerable victim marker do not carry identical meaning across records management systems, and national standards will still need translating into forty-odd data estates.
Lumilinks' work building governed data environments in other regulated sectors, where auditable multi-party sharing is routine, points to a shared semantic layer: agreed definitions, version controlled, tested and mapped onto whatever each force already runs. Effective value delivery via Snowflake means standardisation becomes a modelling exercise deliverable in a quarter rather than a replacement programme measured in years.
Technically this is settled ground. The binding constraint is governance: 43 forces, 43 sets of information-sharing arrangements, moving more slowly than infrastructure ever does. Any supplier who tells you otherwise is glossing over critical details to hide gaps they can't address.
The second challenge has hard numbers behind it. The 2024/2025 MAPPA Annual Report2 put the managed population at 97,050 at 31 March 2025, up 42 per cent in a decade. Licence recall returns to custody at Levels 2 and 3 reached 1,249, a thirteen per cent rise and seventh successive annual increase.
A growing population generating more change events, monitored by teams not growing at the same rate, is a prioritisation problem. Managers hold flat caseload lists reviewed on a cycle, while risk moves daily as individuals pass between custody, probation and community supervision.
What is needed is decision-support capability on trustworthy data foundations:
The method is mature multi-signal scoring, which our team has applied to complex populations for years. Applying it to public protection only changes the context.
Data Protection dictates any model touching individuals, including people who have not reoffended, must be explainable, bias-tested and auditable from the first sprint.
Two principles we hold to without exception:
Handled this way, governance is an enabler. An auditable trail lets a force defend a decision to a court, an inspectorate or a bereaved family, and share data confidently, not cautiously.
Not with a national programme. With a single force or two, one challenge and one measurable outcome: cross-force sharing first as the lower-risk build with the clearest early win, extending into risk scoring only once ethical, legal governance is designed in.
As a specialist UK data and AI consultancy, Lumilinks, combined with Snowflake's enterprise-scale capabilities, helps police forces deliver rapid, measurable outcomes without embarking on another decade-long platform replacement. By building a governed data layer over existing workforce and operational systems, Lumilinks and Snowflake enable a permissioned, zero-copy access across force and partner boundaries, with standardised definitions, full auditability and revocable controls. This creates the governed middle needed to realise the ambitions of the policing white paper and enabling forces to begin delivering value this year.