Local evidence.
Development context.
Each authority brings its own planning context. Dataset scope and development classifications provide the foundation for useful comparisons.
Planning evidence · Refusal Limited
Refusal Atlas
More evidence.
Sharper questions.
Local insight.
The value is in the dataset. Refusal Atlas connects planning refusal reasons, development types and policy references, so individual decisions become a broader picture you can investigate.
DatasetLocal decisions in context
CompareCriteria-led exploration
InvestigateReasons & policy references
A growing evidence base
The dataset is
the foundation.
A single refusal explains one proposal. A collection of reasons reveals which issues recur, how they differ by development type, and where closer investigation is worthwhile.
The collection connects local planning decisions with their development context, covering different types of residential proposals and the refusal themes that recur within them.
The research uses detailed, study-specific classifications to preserve the substance of refusal reasons, rather than reduce every decision to a generic score.
Interface preview
Choose the question.
Set the comparison.
Select a development type and compare two authorities. Switch between counts and shares to explore the interface.
Barking and Dagenham · selected records240Single new dwelling
Waltham Forest · selected records360Single new dwelling
Refusal categories · Single new dwelling
Barking and DagenhamWaltham Forest
Compare category counts within the selected scope. Category shares are different from overall application refusal rates.
LOCAL AREA
Refusal Area boundary
Select a numbered marker to inspect a case. Filter by refusal theme to focus the map.
RESEARCH THREADS
Amenity & policy
01
Review focus
02
How to interpret it
03
Policy context
Interpret policy references in the date-specific context of each planning decision.
Official reference library
National Planning Policy Framework ↗Town and Country Planning Act 1990 ↗These official resources may display newer versions. Use the edition relevant to the decision date when investigating a policy reference.
Dataset depth × geographical breadth
Local detail.
A national research lens.
Each authority adds more than a point on a map. It adds decisions, local policy context and another basis for comparison.
Local evidence.
Find the relevant
comparison set.
The interface design uses searchable authority selection, regional groups and coverage summaries. Users narrow the collection by geography, decision period, development type and evidence availability before comparing results.
Keep the local
and sector context.
Residential studies anchor the collection. The coverage model accommodates commercial and industrial segments, each with relevant categories and a visible sample base. Residential findings stay within their original scope.
More data is useful when it stays comparable.
Retain original refusal wording and study-specific categories, then map them to shared themes for wider comparisons. Keep the mapping version, dates and denominators with each result. New area releases extend the evidence base and give returning users more locations and development contexts to investigate.
Cloud architecture / working design
Built around the data.
Queried around the user.
The architecture separates evidence preservation, AI classification and user-selected statistics. Cloud resources scale with document volume, collection breadth and analytical demand.
Source evidence
Object storage holds documents and extracted text. A source manifest records the authority, application reference, decision date, document location and retrieval history.
AI-enriched records
Processing jobs extract reasons and policy references, retain source passages and assign detailed categories. Versioned classifications support both study-specific analysis and cross-authority themes.
Criteria-led queries
A geospatial database and analytical tables support borough, date, development and reason filters. The query layer returns counts, statistical bases and coverage with the result.
THE RESEARCH-TO-QUERY MODELEvidence → reasons
Evidence → reasons
→ comparisons.
Store once, investigate in many ways. The same structured collection underpins the comparative dashboard, location workspace, policy navigation and institutional AI-agent access through MCP.
- Linked data model
- Authority → application → decision → refusal reason → policy citation. Location, development type and source passages connect these records without losing their original context.
- Classification & quality
- Keep original categories alongside shared themes. Track model and taxonomy versions, uncertain results, duplicates and review status so aggregation remains auditable.
- Statistics & spatial analysis
- Partition analytical data by authority and period, index geographical fields, and cache common comparisons. Query-specific denominators prevent a category count from being mistaken for a refusal rate.
- Publication & access
- Versioned dataset releases, a coverage catalogue and subscription permissions connect users to their selected application pool. A common API serves the web interface and the MCP access design.
For evidence-led development decisions
Ask better questions
before committing.
For investors, developers, planners, architects, building surveyors and owners’ development representatives. Also for NGOs and built environment students exploring how planning policy is applied.
A dataset-led membership model.
Recurring membership centres on access to the analysed planning application pool, with tiers organised by development scale. The product connects statistical exploration with the context behind each result.
Technical depth. Professional context.
Development is led by a computer scientist, with input from surveyors and a legal professional specialising in construction projects. We plan to seek venture funding and scale the evidence collection and analytical capacity with a cloud provider.
Explore the evidence
Which comparison
matters to your decision?
Tell us the authorities, development types and refusal themes relevant to your work.