BLOGS

The UK's planning system has long been described as a bottleneck. Not just in political rhetoric, but in real, measurable cost — to developers carrying land for years waiting on decisions, to lenders pricing risk into stretched appraisals, and to a country that needs to build 1.5 million homes in this Parliament. Now, for the first time in a generation, technology is being deployed at scale to address the root cause: the sheer volume of manual, document-heavy work that has paralysed local authority planning teams.

The government's Extract tool, developed with Google DeepMind's Gemini model and unveiled at London Tech Week in June 2025, reached national rollout across all English local planning authorities by Spring 2026. The ambition is complete digitisation of all planning documents by the end of this year. The implications for housing delivery — and for the risk models underpinning development finance — are significant.

The scale of the backlog problem

Before assessing what Extract can do, it is worth understanding the scale of what it is trying to fix.

Analysis of REalyse's planning data for major residential applications in England reveals a troubling trend. In 2023, the average time from application submission to decision was 435 days. By 2024, that had risen to 462 days. In 2025, it reached 470 days — more than 15 months from submission to outcome, against a statutory target of 13 weeks. These are not fringe cases; they represent the average experience for schemes of ten or more units across the country.

Over those same three years, major residential applications put forward an average of roughly 200,000 units per annum. Approval rates have actually been improving — rising from 76.8% in 2023 to 79.3% in 2025 — suggesting that the problem is not primarily one of political will to approve housing. The bottleneck is administrative capacity. REalyse data currently shows over 362,000 residential units sitting in pending or submitted applications across England, representing the direct cost of a system unable to process its own caseload at pace.

At a national level, approximately 350,000 planning applications are submitted in England each year, each requiring manual validation, document checking, and officer review. The government estimates that planning officers collectively spend around 250,000 hours annually on manual document processing alone — time that is not spent making decisions.

What Extract actually does

Extract converts historic planning maps, scanned PDFs, and handwritten records into structured, machine-readable data. In pilot trials across Hillingdon, Nuneaton & Bedworth, and Exeter councils, the tool digitised planning records — including complex maps — in approximately three minutes each, compared to the one to two hours typically required manually. That translates to roughly 100 planning records processed per day, per deployment.

The productivity case is straightforward. The government projects £527 million in annual public sector savings, against a current council spend of around £59.4 million per year on digital planning and housing software. More importantly, freed from the most labour-intensive stages of document handling, planning officers can redirect capacity toward the work that actually requires professional judgement: assessing applications, engaging with applicants, and making decisions.

The tool does not automate the decision itself. It clears the runway. And in a system where delays have become structural rather than exceptional, that distinction matters enormously.

From digitisation to data infrastructure

Extract's longer-term significance extends beyond processing speed. By converting decades of legacy planning documents into standardised, geospatially indexed data, it creates a richer, more queryable planning record — one that platforms like REalyse can incorporate to give developers, investors, and lenders a clearer picture of site history, local precedent, and pipeline context.

Digitised planning records mean that data on historic decisions, refused applications, appeal outcomes, and site constraints become searchable at scale. For a developer underwriting a site, the ability to quickly interrogate what has and has not been approved in a given postcode district — and at what density — is a meaningful improvement in risk assessment. REalyse's planning data already provides this kind of pipeline visibility; a more complete and current underlying record makes those insights more granular and reliable.

The housing delivery equation

The government's target of 1.5 million homes this Parliament is ambitious by any measure. Housing completions in recent years have fallen consistently short of the 300,000 per annum benchmark, and planning delay is one of the most frequently cited structural barriers. That said, planning approval is only the first gate. Build-out rates, viability, infrastructure delivery, and labour supply all shape the gap between consent and completion.

Where Extract — and the broader digital planning programme — is most likely to have near-term impact is in reducing what the development industry calls "planning drag": the cumulative carrying cost of a site during the period between submission and decision. For a 100-unit scheme at average UK residential land and build values, an additional year of planning delay can add hundreds of thousands of pounds in financing costs, inflation exposure, and opportunity cost. Those costs feed directly into whether a scheme is viable at all, or whether it gets shelved.

If Extract helps compress average major application timelines by even two to three months — a plausible near-term outcome — the downstream effect on scheme viability and development starts could be material. REalyse data on planning pipelines increasingly reflects the difference between areas with fast, well-resourced local authorities and those with persistent backlogs; that gap is likely to narrow as Extract reaches full deployment.

Developer risk calculations are starting to shift

Sophisticated developers and their funders are already beginning to price planning risk differently in areas moving toward digital-first workflows. Where a local authority has modernised its data infrastructure and reduced processing times, lenders are more willing to underwrite pre-planning costs, and developers can model exit timelines with greater confidence.

The Planning and Infrastructure Bill, currently progressing through Parliament alongside the Extract rollout, adds further policy weight to the reform agenda. Taken together, these moves represent the most coordinated attempt to restructure the planning system's operational foundations in decades.

Outlook: reform in motion, not revolution complete

Extract is a genuine step forward. The numbers — 250,000 officer hours reclaimed, £527 million in projected savings, and a digitisation deadline of December 2026 — are credible and meaningful. The trend in REalyse data, showing decision times for major residential schemes approaching 16 months, underscores why reform is overdue.

But Propertymark and other industry bodies are right to note that digitisation alone is not a housing policy. Staffing levels, planning fee structures, local plan coverage, and the political economy of development all remain unresolved. Extract improves what planning officers can do with their time; it does not conjure more of them, nor does it resolve the fundamental tensions between local objection and national housing need.

What it does do — and this matters for anyone with capital deployed in UK residential development — is make the planning system more legible, more data-rich, and more capable of being analysed and predicted. In a market where time is money and planning uncertainty has long been priced as a blunt risk premium, that legibility has real economic value.

For investors and developers using tools like REalyse to track planning pipelines, comparable activity, and site viability, a more digitised system means better signal. And in a housing market still grappling with chronic undersupply, better signal is exactly what decisions should be built on.