How Property Value Estimates Actually Work

Three websites give you three different "estimates" for the same house. None of them is lying — they're just doing the same maths on different data. Here's what's really happening, in plain English.

Type an address into Proplo, Domain or realestate.com.au and you'll get an instant dollar figure. That figure comes from an Automated Valuation Model (AVM) — a piece of software, not a person who's walked through the house. Understanding the three-step recipe behind it tells you exactly how much to trust the number.

Step 1: Find recent comparable sales

Every credible AVM starts the same way a human valuer would — by finding comparable sales: homes of a similar type, size and location that have actually changed hands recently. The more genuine comparables nearby, the more confident the estimate. In a busy suburb like Bondi, there might be dozens in the last year; in a quiet regional town, there might be three.

Step 2: Adjust those sales to today

A sale from 18 months ago isn't worth its 18-month-old price today. The model indexes each comparable forward (or back) using the suburb's own price trend. This is where estimates start to diverge — different providers use different trend data and different time windows, so the same house can legitimately land at $1.45M on one site and $1.55M on another.

Step 3: Blend, and express uncertainty

Finally, the model blends the adjusted comparables (and the property's own last sale, if known) into a single figure — and, if it's honest, a range around it. This is the part most portals hide. A single number like "$1,523,000" implies a precision that no automated model has. That's why Proplo always shows a confidence range and tells you how many comparable sales it's based on.

An estimate is only as good as the sales behind it. If a tool won't show you the comparables, it's asking you to trust a black box.

Why "free" doesn't mean "worse"

Domain and realestate.com.au put estimates and full sales history behind logins and upsells. That's a business-model choice, not a data-quality one — the underlying sales records are government open data that anyone can use. Proplo is built entirely on that same public data, which is why we can give it away free and still show our working: every estimate lists the comparable sales it used, right on the page.

When to trust the number — and when not to

• Trust it more when the suburb has lots of recent sales and the property is a "typical" home for the area.
• Trust it less for unusual properties (acreage, heritage, waterfront), thinly-traded suburbs, or right after a market turn.
• Never treat any AVM — Proplo's included — as a formal valuation. For finance or a purchase decision, get a licensed appraisal.