Sixty-one per cent of UK housing providers say their staff lack the data literacy to make a data strategy work (Housing Technology, 2025). Read it again. The barrier is not the software. It is the people, the operating model, and the assumptions about what data is for. The sector keeps buying tools to fix a problem the tools were never going to solve.
This matters now because the regulatory weather has changed. The Social Housing (Regulation) Act 2023, the consumer standards, Awaab's Law, and the Building Safety Act 2022 each require evidence on demand. Boards are being asked to prove things they do not have the data to prove.
The tooling fallacy
Most data programmes in housing start in the wrong place. A dashboard goes in. A data warehouse gets scoped. A consultant runs a cleansing exercise. Quality improves for a quarter, drift returns, and the same defects reappear in the next reporting cycle.
The reason is structural. Data issues are rarely caused by data alone. They originate upstream in strategy, culture, roles, processes, or system design. Treat the symptom and the symptom comes back. The sector evidence is consistent with this. Thirty-nine per cent of housing providers review their data management processes less than once a year (Housing Technology, 2025). Frameworks exist on paper and decay in practice.
The Housing Technology research surfaces a more uncomfortable finding. Eighty-two per cent of providers have implemented or planned a data ownership framework, yet 34 per cent do not believe their framework is driving improvement (Housing Technology, 2025). The frameworks are in place. The outcomes are not. Something between intent and execution is missing, and it is not technology.
Four levers, one operating system
Strong data maturity rests on four interdependent levers. Strategy sets what data is for. Culture sets whether accuracy and evidence matter in daily work. People hold ownership and stewardship. Business process determines whether data is created and checked correctly at source. Pull one lever in isolation and the others compensate against you.
Strategy is the easiest to get wrong because it is the easiest to declare. A provider under enforcement action does not need the same data outcomes as one preparing for stock transfer growth. Yet most data strategies read identically. They list domains, name a steward, and promise an analytics platform within eighteen months. They do not connect to the regulatory posture, the risk appetite, or the executive accountability the board is willing to enforce.
Culture is harder. Culture shows up in whether a housing officer logs a damp report accurately at first contact, or works around the form because the form is slower than the workaround. Culture is whether the asset team trusts the repairs data enough to use it in capital planning. No dashboard fixes either of these. Trust is built through delivery, not through training campaigns labelled "data literacy" (Batchelder, 2024). The phrase itself signals the problem. Treating data as a separate competence rather than a normal part of operational work tells staff it belongs to someone else.
People are the lever most often confused with technology. Naming a Chief Data Officer does not create accountability. Accountability lives where the consequences of poor data fall on the same person who controls the upstream process. If the housing officer who captured the wrong tenancy start date is not the person who feels the consequence in the void report, the data will keep being wrong.
Business process is where the four levers meet. Validation at the point of capture, single sources of truth for statutory returns, lineage surviving a change of system, and root-cause workflows fixing defects upstream rather than in a Friday afternoon cleansing run. None of this is glamorous. All of it is what produces data a board would defend in front of an inspector.
AI will not rescue you
The sector is talking about AI. Seventy-six per cent of housing providers identify improved data quality as the most important AI-driven development for their organisation, with security and integration close behind (Housing Technology, 2025). Read the result carefully. Providers are not asking AI to deliver predictions yet. They are asking AI to fix the problem they should have fixed before considering AI at all.
Predictive analytics for rent arrears, repairs demand, damp and mould risk, and tenancy sustainment are credible use cases. They are also useless without standardised data models, defined quality thresholds, governed architecture, and executive oversight of data health. A predictive model trained on inconsistent stock condition data produces inconsistent predictions. The model is not the problem. The substrate is.
The harder truth is this. AI exposes the operating model. A working predictive repairs model depends on every preceding handoff being clean: the original asset record, the inspection log, the contractor return, the customer report. When any one of those is unreliable, the prediction fails in a way more visible than a bad spreadsheet ever was. AI is an amplifier. It amplifies what is already there.
What this means for the next twelve months
The providers who will use data well in 2027 are the ones doing unglamorous work in 2026. Fixing capture rules, naming owners with real consequences, certifying a small number of reports and removing the duplicates, embedding data tasks into ordinary operational training rather than parallel literacy programmes, and connecting the strategy to the regulatory posture rather than to a generic maturity ladder.
The dashboard does not save you. The AI does not save you. The four levers, pulled together over time, do.