Accuracy & methodology
Every Victorian zone we returned in our April 2026 benchmark matched the planning scheme. The handful of checks below 100% are reported honestly, and they fail conservative, never silent.
Feasibly is schedule-deep for Victoria: 620+ council zone schedules interpreted from the source planning scheme, machine-verified, and monitored daily for amendments.
The benchmark
Sampled across residential and commercial sites — CBD, activity centres, suburban streets, each value checked against the authoritative source. Nothing rounded up.
| Check | Result | How it was measured |
|---|---|---|
| Zone — residential | 100% · 20/20 | Domain sold properties vs VicPlan WFS at the geocoded point. |
| Zone — commercial | 100% · 20/20 | CBD, activity centres and commercial strips vs VicPlan WFS. |
| Zone schedule values | 100% · 620/620 | Every council schedule value (min lot, garden area, height, setbacks) Playwright-verified against the planning scheme website. |
| ResCode (Clause 55) formulas | 100% | Building-envelope outputs vs the official DTP PPN27 worked examples. |
| Overlays | 95% exact | Boundary-intersection filtering vs VicPlan WFS. The other 5% return a superset — we flag a boundary overlay rather than miss one. Conservative, never silent. |
| Lot area | 90% · 18/20 | Cadastre geometry vs Domain-listed size; the gaps are listing-vs-title rounding, not zone error. |
The wrong-zone guarantee is staked on the line that decides a screening: the zone. That’s 100% on this benchmark, and if it’s ever wrong for you, the lookup is free and we credit ten more, with the fix published in the changelog.
How we get there
Accuracy isn’t a one-time score — it’s a process that keeps the data right as the schemes change underneath it.
For Victoria we don’t just read a spatial layer — we interpret 620+ council-specific zone schedules from the source planning scheme. The minimum lot size, garden area, height and setbacks you get are the council’s actual numbers, not a state default.
A Playwright verifier opens the planning scheme website and checks each schedule value against the published clause. 620 of 620 matched at the last benchmark, and it re-runs as schedules change.
The 2,273 overlay-permit schedules behind “does this need a permit?” are each extracted and then confirmed by a three-judge majority vote across independent models. Single-model guesses don’t enter the dataset; anything unverified is labelled as such.
A scheme monitor checks the Victorian planning schemes every morning. When a council gazettes an amendment, it’s flagged, and any correction lands in the public changelog with date, council and source URL.
Each value is tagged verified_schedule, machine-extract or not-codified, with the source URL. You always know whether a number is human-verified or auto-extracted — the honesty is in the response, not just the marketing.
Coverage isn’t uniform, and we won’t pretend it is. Victoria is the schedule-deep tier above. New South Wales, South Australia, Queensland (SEQ), Western Australia, the ACT and Tasmania are API-deep — zone and controls come straight from each government’s authoritative spatial API (NSW, SA and QLD also benchmarked at 100% zone), without the per-council schedule interpretation Victoria gets. Both are honest tiers; we never blur them.
Run a free lookup and read the confidence tier and source URL on every value.
Wrong zone? Free lookup + 10 credits + a public correction. Every time.