About PropertyCheck

PropertyCheck exists because Australian property data is either locked behind expensive subscriptions or served up as a black-box number you're asked to trust. We publish the estimate and the evidence — free, on every property.

How our estimates work

1. Comparable selection. We gather recent sales of similar dwellings in the same suburb — matched on dwelling type, bedrooms, bathrooms, parking and land or floor size, weighted by similarity and recency.
2. Time indexing. Each sale price is brought forward to the current month using the suburb's dwelling-type price index, so a sale from last spring is compared in today's dollars.
3. Attribute adjustment. A hedonic model adjusts for differences — an extra bedroom, a smaller block, a renovated era — using coefficients estimated from the whole market.
4. Blending & anchoring. The comparable evidence is blended with a hedonic baseline, and when the property itself sold recently, that sale anchors the estimate strongly — real transactions beat models.
5. Honest uncertainty. The published range is the model's deviation across the evidence. Confidence is High only with dense, tight comparable evidence; unusual homes are labelled low confidence on purpose.

What an estimate is not

An automated estimate is not a professional valuation and is never a substitute for one where money or law is involved — lending decisions, family settlements, taxation. For those, engage a licensed valuer. Before selling, get an in-person appraisal from a local agent — we'll match you free.

Data sources

This demonstration build runs on a clearly-labelled synthetic dataset. The production architecture is designed for: G-NAF (Australia's open geocoded address file), state land-registry sales records (licensed per state), and ABS Census demographics — with every figure traceable to its source.