Data Methodology

Overview

One engine powers every PlotDetect product — the Intelligence Brief, Verify, the Conveyancer report, the site reports, and the free Scout and Validate tools. This page describes how each kind of data behind them is produced, checked, maintained, and delivered.

The engine draws on four data layers:

1. Live NSW Planning Portal queries — zoning, standards, hazards and lot geometry, fetched at the moment of each search (all products)
2. The structured planning corpus — DCP, LEP and SEPP provisions extracted into a database, each cited to its source clause and page (Verify, the Brief, the Conveyancer report)
3. Government datasets — land values, strata records, development application data and mapped constraint layers
4. Satellite and terrain analysis — the property observed from imagery and elevation data (the site reports and Brief add-ons)

Source Documents

PlotDetect's primary sources are the official planning instruments published by NSW local councils and the NSW Government:

• Development Control Plans (DCPs): Detailed planning controls covering setbacks, heights, floor space ratios, heritage provisions, landscaping, parking, and site-specific requirements
• Local Environmental Plans (LEPs): Statutory instruments defining zoning, land use permissibility, heritage listings, and development standards
• State Environmental Planning Policies (SEPPs): State-level instruments including SEPP Housing 2021, SEPP Exempt and Complying Development Codes 2008, and others — sourced directly from the NSW Legislation website
• Heritage schedules and conservation area maps: Council-gazetted heritage items, heritage conservation areas, and associated provisions
• NSW Planning Portal (live): Property zoning, heritage overlays, flood and bushfire prone land, FSR and height limits, and lot geometry are queried in real time from the NSW Planning Portal spatial API at the moment of each search — not from a cached copy. This is the same authoritative spatial data used by NSW councils and the Department of Planning

All source documents are publicly available from council websites and the NSW Legislation website. PlotDetect does not use proprietary or non-public data sources for planning provisions.

Extraction Process

Planning provisions are extracted from source PDFs through a structured process:

1. Document parsing: DCP and LEP documents are parsed to identify individual provisions, their hierarchical structure (part, section, clause, sub-clause), and relationships to zones, precincts, and development types. Where AI assists in reading a document, every extracted value is checked against the source text and human-reviewed before it enters the database
2. Provision structuring: Each extracted provision is stored with structured metadata including:
   • Provision text (verbatim from source document)
   • Source document name and version
   • Section, clause, and sub-clause references
   • PDF page number
   • Applicable zones and land use categories
   • Applicable precincts (where spatially defined)
   • Development types the provision applies to
   • Whether the provision is a numerical standard, design guideline, or heritage-specific control
3. Spatial indexing: Precinct boundaries are digitised and stored as PostGIS geometries, enabling spatial queries that match a property location to its applicable precinct-specific provisions
4. Relationship mapping: Cross-references between provisions (e.g., "see also Section 4.3", "subject to the requirements of Part 6") are captured as structured relationships, not loose text

Automated Pathway Triage

PlotDetect runs automated checks across the approval pathways available for a property, presenting the fastest pathway first:

Traceable by design: Pathway triage is computed from published SEPP provisions using deterministic rules, not AI interpretation. Every trigger checked and every exclusion identified can be traced to its source clause and PDF page in the legislation.

Provenance Chain

Every provision in PlotDetect's database maintains a complete provenance chain:

This chain means any provision returned by PlotDetect can be independently checked by a user opening the source DCP document and navigating to the cited page and section.

Data Currency

Planning instruments are amended by councils periodically. PlotDetect tracks data currency at the council level:

• Each council's DCP has a recorded "last parsed" date, indicating when PlotDetect last extracted provisions from the current version
• When a council publishes an amended DCP, PlotDetect re-parses the updated document and flags changed, added, or removed provisions
• Data currency dates are disclosed to users so they can assess whether the data reflects the most current planning instrument

There may be a lag between when a council amends a DCP and when PlotDetect reflects the change. Users should always verify critical decisions against the current version of the source document, available from the relevant council's website.

Heritage Data

Heritage provisions receive additional structured treatment:

• Heritage item identification: Properties listed as heritage items under the LEP are identified with their item number, name, address, and significance level (local or state)
• Heritage conservation areas: Properties within heritage conservation areas (HCAs) are identified with the applicable HCA name and boundary
• Heritage-specific DCP provisions: Provisions from heritage-specific sections of DCPs are linked to the relevant heritage items and HCAs, including controls on materials, scale, setbacks, demolition, and alterations
• Provision chains: Where heritage provisions reference or modify general provisions (e.g., "despite Section 3.2, heritage items in the Haberfield HCA must..."), these override relationships are captured

Quality Assurance

• Automated validation: Extracted provisions are validated for completeness (mandatory fields populated), consistency (zone codes match LEP definitions), and referential integrity (cross-references resolve)
• Sample verification: A random sample of provisions is manually checked against source documents after each extraction run
• User feedback: Reported inaccuracies are investigated, corrected, and logged
• Mapped-area (precinct) boundaries: Some DCP chapters apply only inside areas defined by map figures in the adopted plan. Before such a boundary is used for address matching, it passes geometry checks, geocoded point checks against the adopted figure (including addresses that must not match), and provenance recording; where a boundary has not passed these checks, the area's rules are excluded and the report discloses that site-specific controls may exist. The full method is described in How It Works

Government Datasets

Beyond the planning instruments, PlotDetect queries public government datasets and reports what they return, with the source and retrieval date attached:

• Mapped constraint layers: flood, bushfire, heritage, acid sulfate, biodiversity, riparian, wetlands, coastal and related planning layers from NSW Government spatial services. Results list which layers were checked for that council, so gaps in mapping are visible rather than silent
• Land values: NSW Valuer General land values, base dates and valuation history
• Strata records: NSW Strata Hub data used to identify whether a lot is strata or Torrens title
• Development applications: DA and CDC records from the NSW Planning Portal — used for nearby-application listings and for the DA analytics in Validate, which cover roughly 98 councils

Satellite & Terrain Analysis

The site reports (flood, bushfire, solar, granny flat, overshadowing, development monitoring) and the Brief's satellite add-ons observe the property itself, from public imagery and elevation data:

• Sources: open satellite imagery programs (such as the European Space Agency's Sentinel missions), NSW Government aerial imagery, and public elevation models
• Measured vs modelled — always labelled: some outputs are observations (structures and vegetation visible in imagery, terrain slope and elevation); others are model outputs (shadow geometry computed from the planning height envelope, solar potential estimated from roof geometry and location, flood screening combining terrain with satellite observations). Model outputs are labelled as estimates and are never presented as measurements or as published planning values
• Change over time: development monitoring compares satellite scenes of an area across time and flags where the surface has changed — it detects that something changed, not what was built

How the Intelligence Brief Assembles It

The Intelligence Brief and the Conveyancer report draw on all four layers for one address. Two rules govern the assembly:

• Every value carries its provenance: the source (instrument, dataset or model), the date it was retrieved, and a label distinguishing published values, extracted provisions and model estimates
• Gaps are stated, not filled: where a source is unavailable, not mapped for that council, or times out, the section says so and gives the reason — a missing answer is reported as missing, never replaced with a guess

What PlotDetect Does Not Do

• PlotDetect does not interpret provisions or assess whether a specific proposal complies
• PlotDetect does not use generative AI to answer queries — results are database lookups of structured provisions. Where AI assists in reading source documents during extraction, every extracted value is checked against the source text and human-reviewed before it enters the database, and carries its clause and page citation (see our AI Disclaimer)
• PlotDetect does not provide legal advice or replace the role of qualified town planners, heritage consultants, or legal professionals
• PlotDetect does not guarantee completeness — some provisions may involve subjective criteria, site-specific conditions, or recent amendments not yet reflected in the database

Coverage

Coverage is layered, and each product shows the depth available for the address you enter:

• Any NSW address: zoning, SEPP and LEP standards, hazard checks and DA data — live from the NSW Planning Portal
• Structured numeric DCP controls (setbacks, parking, landscaping and similar) across 24 Sydney councils
• The deepest full-DCP integration is currently in the Inner West, with several eastern and northern councils holding partial provision text

The hand-built council layer expands deepest-first across metropolitan Sydney — prioritised by transaction density, property value and planning complexity (see the coverage section on the homepage). Data currency dates for a specific council are available on request: [email protected].

Related Pages

• AI Disclaimer & FAQ — how deterministic data retrieval works, what AI is and is not used for, and answers to common questions from planning professionals and government stakeholders
• Audit Trail — query logging, data version history, retention periods, and NSW AIAF alignment
• How It Works — per-tool data sources, update cadence, limitations, and where location-specific (precinct) rules come from