Market Insights11 min read
How Properties Are Valued: Methods, Myths, and What AI Actually Adds
PA
PropertyLens AI## The Number Everyone Argues About
When a Paddington Queenslander sells for $1.72 million and the identical house two streets over sold for $1.45 million eight months earlier, someone has to explain the gap. Was the first buyer reckless? Was the second a bargain? Or does each property carry a value that no single method can fully capture?
Property valuation is one of the most consequential calculations in personal finance, yet most buyers and sellers have only a vague sense of how it works. A bank valuer, a real estate agent, an automated model, and a buyer at auction can all look at the same property and arrive at genuinely different numbers — and all of them can be right, depending on what question they're trying to answer.
Here's how each method works, what it's actually measuring, and where the gaps appear.
## The Four Main Valuation Methods
### 1. Comparable Sales (Direct Comparison)
This is the backbone of residential property valuation in Australia. The logic is straightforward: a property is worth what a willing buyer will pay a willing seller in an open market, and recent sales of similar properties are the best evidence of that.
A valuer using this method identifies three to six comparable sales — ideally within the same suburb, sold within the past three to six months, with similar land size, dwelling size, condition, and configuration. They then make adjustments for differences: an extra bathroom might add $30,000–$50,000 in Bulimba; a main road frontage might subtract 8–12% in Annerley.
The method works well when there are enough comparable sales to draw on. It struggles in tightly held streets where only two or three properties have sold in the past year, or when a property is genuinely unusual — a 900sqm corner block in Ascot with a heritage-listed home doesn't have many true comparables.
**The myth**: Many sellers believe their renovation automatically adds dollar-for-dollar value. It rarely does. A $120,000 kitchen and bathroom renovation in a suburb with a $900,000 median might add $60,000–$80,000 to the comparable sales assessment, not $120,000. The market sets the ceiling, not the renovation cost.
### 2. Capitalisation Rate (Income Approach)
Used primarily for investment properties, this method values a property based on the income it generates. The formula:
**Value = Net Operating Income ÷ Capitalisation Rate**
If a Fortitude Valley apartment generates $28,000 per year in net rent, and comparable investment properties in that precinct are trading at a 4.5% capitalisation rate, the implied value is approximately $622,000.
The capitalisation rate (or cap rate) is derived from comparable investment sales — it reflects what investors are currently willing to accept as a return for that asset class in that location. In inner Brisbane, residential cap rates have been running between 3.5% and 5.5% depending on property type and suburb, with commercial assets carrying different benchmarks entirely.
This method is most useful when comparing investment options or assessing whether a price is justified by income. It's less relevant for owner-occupied residential property, where emotional value and lifestyle factors drive a significant portion of price.
**The myth**: A high rental yield always means a good investment. Not necessarily. A 6.5% gross yield in a suburb with flat or declining capital growth may underperform a 3.8% yield property in a suburb growing at 7–9% annually. The income approach captures current yield, not total return.
### 3. Summation Method (Cost Approach)
This method adds up the components: land value plus the depreciated replacement cost of improvements (the dwelling and any structures).
**Value = Land Value + Depreciated Building Value**
For a post-war house in Moorooka on a 607sqm block, you might estimate the land at $680,000 (based on recent vacant land or land-only sales), add the replacement cost of the dwelling at $350,000 (roughly $2,800/sqm for a modest 125sqm home), then apply depreciation for age and condition — say 25% — arriving at a building value of $262,500 and a total of approximately $942,500.
This method is most commonly used by insurance valuers, quantity surveyors, and in situations where there are few comparable sales — new construction, unusual properties, or rural holdings. It's also used by the Queensland Valuer-General when setting unimproved land values for rating and land tax purposes.
For standard residential properties in inner Brisbane, the summation method often produces values below market because it doesn't capture the scarcity premium, the desirability of a location, or what buyers will pay above replacement cost to live in a particular suburb.
**The myth**: If you can build it for less than the asking price, you're overpaying. This ignores location value entirely. You cannot build a house in Teneriffe for $800,000 and have it sit on a Teneriffe block — the land alone would cost more than that.
### 4. Hypothetical Development (Residual) Method
Used for development sites and properties with subdivision or redevelopment potential. The valuer estimates the gross realisation of the completed development, subtracts all costs (construction, finance, holding costs, margins), and works backwards to determine what a developer could pay for the land.
This method is highly sensitive to assumptions. Change the construction cost estimate by 5% or the assumed sale price by 3% and the residual land value can shift by 15–20%. It's why development site valuations carry wider confidence intervals than standard residential valuations.
For a Brisbane homeowner wondering if their 800sqm corner block in Nundah has development value, this is the method a town planner or feasibility consultant would use — and it requires current build cost data, current comparable new product sales, and a realistic assessment of what the planning scheme actually permits.
## Bank Valuations vs Market Value: The Gap That Catches Buyers Out
This is where many buyers get a shock. A bank valuation is not the same as market value, and the difference can be significant.
A bank valuation is a **mortgage security valuation** — it answers the question: if we had to sell this property quickly in a distressed scenario, what would we confidently recover? Banks instruct registered valuers to be conservative. They're protecting their loan book, not validating your purchase price.
In a rising market, bank valuations routinely come in 5–10% below the contract price on auction purchases. A buyer who paid $1.38 million at auction in Coorparoo might receive a bank valuation of $1.26 million. If they're borrowing at 80% LVR, the bank will lend against $1.26 million, not $1.38 million — meaning the buyer needs to find an additional $96,000 in cash or equity.
This gap tends to widen during periods of rapid price growth. In 2021–2022, when Brisbane prices were rising 2–3% per month in some suburbs, bank valuations were consistently lagging the market because valuers rely on settled sales data, which is inherently backward-looking.
The practical implication: if you're buying at auction with a tight deposit, get a pre-auction valuation or build a buffer into your borrowing capacity calculations.
## Automated Valuation Models: What They Do and Don't Do
Automated Valuation Models (AVMs) have been around since the early 2000s. The major property data providers — CoreLogic, PropTrack, Domain — all run AVMs that produce instant estimates for most Australian residential properties.
AVMs work by applying statistical algorithms to large datasets: recent sales, property characteristics (bedrooms, bathrooms, land size, building size), suburb price trends, and distance-based adjustments. They're fast, cheap, and reasonably accurate at the median — CoreLogic's AVM claims a median error rate of around 5–7% for properties where there's sufficient sales data.
But the limitations are real:
- **They can't see inside the property.** A renovated Queenslander and an unrenovated one on the same street will have similar AVM estimates if the algorithm can't distinguish condition.
- **They struggle with outliers.** Unusual properties — large blocks, unusual configurations, heritage listings, properties with significant views — sit outside the normal distribution that AVMs are trained on.
- **They lag the market.** AVMs are trained on settled sales, which reflect contracts signed 30–90 days earlier. In a fast-moving market, they can be materially behind.
- **Thin data suburbs produce wide confidence intervals.** A suburb with 15 sales per year gives an AVM far less to work with than one with 150.
The AVM estimate you see on a real estate portal is a starting point for a conversation, not a valuation.
## What AI-Powered Prediction Actually Does Differently
There's a meaningful difference between a traditional AVM and a machine learning model trained specifically on a defined market with curated data.
PropertyLens's AI price prediction tool is built on Brisbane sales data and trained to recognise patterns that traditional AVMs miss — the premium attached to specific street orientations in Hawthorne, the discount applied to ground-floor units on arterial roads in Spring Hill, the way proximity to a new bus rapid transit stop has repriced properties in Woolloongabba over the past 18 months.
The key differences from a standard AVM:
- **Granularity**: Rather than suburb-level adjustments, the model works at a street and property-characteristic level for inner Brisbane, where it has the deepest data.
- **Feature weighting**: Machine learning models can identify which features actually drive price in a given suburb — which isn't always what you'd expect. In some inner Brisbane suburbs, land size matters more than dwelling size; in others, the relationship reverses.
- **Transparency**: PropertyLens shows confidence intervals alongside predictions. A property with many close comparables will have a tight range; an unusual property will show a wider one. That honesty matters.
- **Recency**: The model is updated as new sales data flows through, reducing the lag that affects traditional AVMs.
What it doesn't do: replace a registered valuer for mortgage purposes, account for information only visible on inspection, or predict what a specific emotional buyer might pay on a specific auction day. No model does that.
## Practical Implications for Buyers and Sellers
**If you're buying:**
- Use comparable sales as your primary anchor. Pull the last 12 months of sales in the target suburb, filter for similar properties, and build your own range before you look at any automated estimate.
- Understand that the agent's price guide is a marketing tool, not a valuation. In Queensland, agents are required to provide a price range, but the range is set with vendor expectations and buyer attraction in mind.
- If you're borrowing close to your limit, get a pre-purchase valuation from a registered valuer ($300–$600) before bidding at auction. It's cheap insurance.
- AI prediction tools can help you sense-check whether a property is priced within a reasonable range for its characteristics — useful for narrowing your research before committing to deeper due diligence.
**If you're selling:**
- Get at least two agent appraisals and ask each agent to show you the comparable sales they used. If they can't produce them, the appraisal is guesswork.
- Understand the difference between an appraisal (an agent's opinion, not regulated) and a formal valuation (produced by a registered valuer under professional standards).
- Renovation value is real but bounded by suburb medians. In a suburb with a $1.1 million median, a $200,000 renovation rarely produces a $1.3 million sale price.
- Timing matters. A property valued in February 2026 in a suburb with strong clearance rates will likely attract a different result than the same property listed in a softer market.
**If you're refinancing or accessing equity:**
- The bank's valuer is working for the bank, not for you. If the valuation comes in low, you can request a review with additional comparable sales evidence — it sometimes works.
- In a rising market, waiting 6–12 months before refinancing can make a material difference to the bank valuation, as more recent comparable sales flow through.
## The Bottom Line
Property valuation is not a single number — it's a range, and the right method depends on what question you're trying to answer. Comparable sales tell you what the market has paid. The income approach tells you what an investor should pay. The summation method tells you what it would cost to replicate. The bank valuation tells you what a lender will secure against. And AI prediction tools tell you where the data suggests value sits, with appropriate uncertainty attached.
The buyers and sellers who make the best decisions are the ones who understand all four methods, know their limitations, and triangulate rather than anchoring on any single figure.
PropertyLens's Market Dashboard and AI Price Predictions are built to support exactly that kind of triangulation — pulling together comparable sales data, suburb trends, and machine learning estimates for inner Brisbane properties in one place. Worth exploring before your next decision, whether you're buying, selling, or simply trying to understand what you already own.
When a Paddington Queenslander sells for $1.72 million and the identical house two streets over sold for $1.45 million eight months earlier, someone has to explain the gap. Was the first buyer reckless? Was the second a bargain? Or does each property carry a value that no single method can fully capture?
Property valuation is one of the most consequential calculations in personal finance, yet most buyers and sellers have only a vague sense of how it works. A bank valuer, a real estate agent, an automated model, and a buyer at auction can all look at the same property and arrive at genuinely different numbers — and all of them can be right, depending on what question they're trying to answer.
Here's how each method works, what it's actually measuring, and where the gaps appear.
## The Four Main Valuation Methods
### 1. Comparable Sales (Direct Comparison)
This is the backbone of residential property valuation in Australia. The logic is straightforward: a property is worth what a willing buyer will pay a willing seller in an open market, and recent sales of similar properties are the best evidence of that.
A valuer using this method identifies three to six comparable sales — ideally within the same suburb, sold within the past three to six months, with similar land size, dwelling size, condition, and configuration. They then make adjustments for differences: an extra bathroom might add $30,000–$50,000 in Bulimba; a main road frontage might subtract 8–12% in Annerley.
The method works well when there are enough comparable sales to draw on. It struggles in tightly held streets where only two or three properties have sold in the past year, or when a property is genuinely unusual — a 900sqm corner block in Ascot with a heritage-listed home doesn't have many true comparables.
**The myth**: Many sellers believe their renovation automatically adds dollar-for-dollar value. It rarely does. A $120,000 kitchen and bathroom renovation in a suburb with a $900,000 median might add $60,000–$80,000 to the comparable sales assessment, not $120,000. The market sets the ceiling, not the renovation cost.
### 2. Capitalisation Rate (Income Approach)
Used primarily for investment properties, this method values a property based on the income it generates. The formula:
**Value = Net Operating Income ÷ Capitalisation Rate**
If a Fortitude Valley apartment generates $28,000 per year in net rent, and comparable investment properties in that precinct are trading at a 4.5% capitalisation rate, the implied value is approximately $622,000.
The capitalisation rate (or cap rate) is derived from comparable investment sales — it reflects what investors are currently willing to accept as a return for that asset class in that location. In inner Brisbane, residential cap rates have been running between 3.5% and 5.5% depending on property type and suburb, with commercial assets carrying different benchmarks entirely.
This method is most useful when comparing investment options or assessing whether a price is justified by income. It's less relevant for owner-occupied residential property, where emotional value and lifestyle factors drive a significant portion of price.
**The myth**: A high rental yield always means a good investment. Not necessarily. A 6.5% gross yield in a suburb with flat or declining capital growth may underperform a 3.8% yield property in a suburb growing at 7–9% annually. The income approach captures current yield, not total return.
### 3. Summation Method (Cost Approach)
This method adds up the components: land value plus the depreciated replacement cost of improvements (the dwelling and any structures).
**Value = Land Value + Depreciated Building Value**
For a post-war house in Moorooka on a 607sqm block, you might estimate the land at $680,000 (based on recent vacant land or land-only sales), add the replacement cost of the dwelling at $350,000 (roughly $2,800/sqm for a modest 125sqm home), then apply depreciation for age and condition — say 25% — arriving at a building value of $262,500 and a total of approximately $942,500.
This method is most commonly used by insurance valuers, quantity surveyors, and in situations where there are few comparable sales — new construction, unusual properties, or rural holdings. It's also used by the Queensland Valuer-General when setting unimproved land values for rating and land tax purposes.
For standard residential properties in inner Brisbane, the summation method often produces values below market because it doesn't capture the scarcity premium, the desirability of a location, or what buyers will pay above replacement cost to live in a particular suburb.
**The myth**: If you can build it for less than the asking price, you're overpaying. This ignores location value entirely. You cannot build a house in Teneriffe for $800,000 and have it sit on a Teneriffe block — the land alone would cost more than that.
### 4. Hypothetical Development (Residual) Method
Used for development sites and properties with subdivision or redevelopment potential. The valuer estimates the gross realisation of the completed development, subtracts all costs (construction, finance, holding costs, margins), and works backwards to determine what a developer could pay for the land.
This method is highly sensitive to assumptions. Change the construction cost estimate by 5% or the assumed sale price by 3% and the residual land value can shift by 15–20%. It's why development site valuations carry wider confidence intervals than standard residential valuations.
For a Brisbane homeowner wondering if their 800sqm corner block in Nundah has development value, this is the method a town planner or feasibility consultant would use — and it requires current build cost data, current comparable new product sales, and a realistic assessment of what the planning scheme actually permits.
## Bank Valuations vs Market Value: The Gap That Catches Buyers Out
This is where many buyers get a shock. A bank valuation is not the same as market value, and the difference can be significant.
A bank valuation is a **mortgage security valuation** — it answers the question: if we had to sell this property quickly in a distressed scenario, what would we confidently recover? Banks instruct registered valuers to be conservative. They're protecting their loan book, not validating your purchase price.
In a rising market, bank valuations routinely come in 5–10% below the contract price on auction purchases. A buyer who paid $1.38 million at auction in Coorparoo might receive a bank valuation of $1.26 million. If they're borrowing at 80% LVR, the bank will lend against $1.26 million, not $1.38 million — meaning the buyer needs to find an additional $96,000 in cash or equity.
This gap tends to widen during periods of rapid price growth. In 2021–2022, when Brisbane prices were rising 2–3% per month in some suburbs, bank valuations were consistently lagging the market because valuers rely on settled sales data, which is inherently backward-looking.
The practical implication: if you're buying at auction with a tight deposit, get a pre-auction valuation or build a buffer into your borrowing capacity calculations.
## Automated Valuation Models: What They Do and Don't Do
Automated Valuation Models (AVMs) have been around since the early 2000s. The major property data providers — CoreLogic, PropTrack, Domain — all run AVMs that produce instant estimates for most Australian residential properties.
AVMs work by applying statistical algorithms to large datasets: recent sales, property characteristics (bedrooms, bathrooms, land size, building size), suburb price trends, and distance-based adjustments. They're fast, cheap, and reasonably accurate at the median — CoreLogic's AVM claims a median error rate of around 5–7% for properties where there's sufficient sales data.
But the limitations are real:
- **They can't see inside the property.** A renovated Queenslander and an unrenovated one on the same street will have similar AVM estimates if the algorithm can't distinguish condition.
- **They struggle with outliers.** Unusual properties — large blocks, unusual configurations, heritage listings, properties with significant views — sit outside the normal distribution that AVMs are trained on.
- **They lag the market.** AVMs are trained on settled sales, which reflect contracts signed 30–90 days earlier. In a fast-moving market, they can be materially behind.
- **Thin data suburbs produce wide confidence intervals.** A suburb with 15 sales per year gives an AVM far less to work with than one with 150.
The AVM estimate you see on a real estate portal is a starting point for a conversation, not a valuation.
## What AI-Powered Prediction Actually Does Differently
There's a meaningful difference between a traditional AVM and a machine learning model trained specifically on a defined market with curated data.
PropertyLens's AI price prediction tool is built on Brisbane sales data and trained to recognise patterns that traditional AVMs miss — the premium attached to specific street orientations in Hawthorne, the discount applied to ground-floor units on arterial roads in Spring Hill, the way proximity to a new bus rapid transit stop has repriced properties in Woolloongabba over the past 18 months.
The key differences from a standard AVM:
- **Granularity**: Rather than suburb-level adjustments, the model works at a street and property-characteristic level for inner Brisbane, where it has the deepest data.
- **Feature weighting**: Machine learning models can identify which features actually drive price in a given suburb — which isn't always what you'd expect. In some inner Brisbane suburbs, land size matters more than dwelling size; in others, the relationship reverses.
- **Transparency**: PropertyLens shows confidence intervals alongside predictions. A property with many close comparables will have a tight range; an unusual property will show a wider one. That honesty matters.
- **Recency**: The model is updated as new sales data flows through, reducing the lag that affects traditional AVMs.
What it doesn't do: replace a registered valuer for mortgage purposes, account for information only visible on inspection, or predict what a specific emotional buyer might pay on a specific auction day. No model does that.
## Practical Implications for Buyers and Sellers
**If you're buying:**
- Use comparable sales as your primary anchor. Pull the last 12 months of sales in the target suburb, filter for similar properties, and build your own range before you look at any automated estimate.
- Understand that the agent's price guide is a marketing tool, not a valuation. In Queensland, agents are required to provide a price range, but the range is set with vendor expectations and buyer attraction in mind.
- If you're borrowing close to your limit, get a pre-purchase valuation from a registered valuer ($300–$600) before bidding at auction. It's cheap insurance.
- AI prediction tools can help you sense-check whether a property is priced within a reasonable range for its characteristics — useful for narrowing your research before committing to deeper due diligence.
**If you're selling:**
- Get at least two agent appraisals and ask each agent to show you the comparable sales they used. If they can't produce them, the appraisal is guesswork.
- Understand the difference between an appraisal (an agent's opinion, not regulated) and a formal valuation (produced by a registered valuer under professional standards).
- Renovation value is real but bounded by suburb medians. In a suburb with a $1.1 million median, a $200,000 renovation rarely produces a $1.3 million sale price.
- Timing matters. A property valued in February 2026 in a suburb with strong clearance rates will likely attract a different result than the same property listed in a softer market.
**If you're refinancing or accessing equity:**
- The bank's valuer is working for the bank, not for you. If the valuation comes in low, you can request a review with additional comparable sales evidence — it sometimes works.
- In a rising market, waiting 6–12 months before refinancing can make a material difference to the bank valuation, as more recent comparable sales flow through.
## The Bottom Line
Property valuation is not a single number — it's a range, and the right method depends on what question you're trying to answer. Comparable sales tell you what the market has paid. The income approach tells you what an investor should pay. The summation method tells you what it would cost to replicate. The bank valuation tells you what a lender will secure against. And AI prediction tools tell you where the data suggests value sits, with appropriate uncertainty attached.
The buyers and sellers who make the best decisions are the ones who understand all four methods, know their limitations, and triangulate rather than anchoring on any single figure.
PropertyLens's Market Dashboard and AI Price Predictions are built to support exactly that kind of triangulation — pulling together comparable sales data, suburb trends, and machine learning estimates for inner Brisbane properties in one place. Worth exploring before your next decision, whether you're buying, selling, or simply trying to understand what you already own.