A rookie who sees themselves as an expert in our industry says, “You can’t argue with data.” He later boasted about having suburb reports available in minutes on their platform. The fact is, the reports have been available for at least 20 years through CoreLogic’s RPData Product. Their posts on social media are all produced by AI
When Confidence Replaces Competence: The Hidden Risks of Misused Data in Property Buying
When Confidence Replaces Competence emerges, it often appears when shiny platforms replace fundamental principles. Raw data is the anchor. AI and algorithms are only helpful after you verify the facts underlying them. Treat the model as a guide. Let raw data make the call.
Here’s the short answer: rely on raw data. Then use tools to assist you. Use prompts that you know of. A platform cannot show its infrastructure or how it is programmed.
Why are Many Buyer’s Agents Jumping on Platforms
Platforms look efficient. They promise speed, dashboards and a “confidence score.” Many buyers agents now lean on these tools without understanding the training data, what was excluded, or how often the feeds update. That’s where When Confidence Replaces Competence begins.
• If the agent doesn’t understand how the machine was built, they risk making catastrophic mistakes
• If they cannot reproduce a price view from raw sales and rentals, they are trusting what the machine generated.
• If they cannot explain model limits and error rates in plain English, they should not recommend the output or use it.
Consumers bear the risk. A 6% error on a $1.8 million house is $108,000 before costs. Add two or three wasted reports and empty weeks between tenants, and the damage can jump well into the hundreds of thousands over a hold period.
Can AI pick the best property for you?
Not without checking raw data. Models smooth reality and can miss current shifts in the marketplace. Use them to shortlist. Then confirm with very recent comparable sales, days on market, rental absorption and the council pipeline. If those facts do not support the pick, it is out.
What a competent buyer’s agent must know before relying on a platform
Sources and coverage: titles, sales, rentals, planning, overlays, how far back, what’s missing
• Updates: how often the data updates, and lag times by source
• Training and bias: how the model was trained, what features it used, known blind spots
• Error behaviour: typical absolute error, worst case, and where it breaks
• Exclusions and smoothing: outlier policy, “imputations,” and any suburb-level averaging
• Replication: ability to reproduce the platform’s view with raw data from a tight window
• Counter-checks: what would prove the platform wrong for this property, on this street, right now
A buyer’s agent who cannot answer these questions in plain English should not recommend a property based solely on a platform report.
Raw data first: transparency, integrity, freshness
Raw data shows what actually happened. You can see dates, sources and the exact records. That lets you test claims instead of trusting a chart. Freshness matters too. A three-month window often tells you more than a yearly average. Street-level and building-level reads beat suburb medians when markets turn.
Algorithm limits when raw data is missing
Algorithms learn from history. If history is patchy or skewed, the model repeats that bias. Inputs may also be “cleaned” before they go in by dropping outliers or smoothing spikes. Most of all, algorithms miss context. They do not hear aircraft over a new flight path, see a shadowed yard at 3 pm, or feel a street after dark. Raw data plus ground truth fixes that.
Robodebt and property: automated certainty is not proof
Robodebt demonstrated what happens when a system appears authoritative, yet its inputs and logic are flawed. It punished people because no one checked the method against reality. Property tools can fail the same way. When a dashboard says buy here now, could you ask to see the raw inputs and method? If there are any grey areas, flag them. Read the public record yourself: https://www.robodebt.royalcommission.gov.au
Why Suburb Medians Prices Will Not Save You
A median hides detail. Two streets can move in opposite directions as a new building is constructed, a flight path shifts, or a significant event occurs. Read street-level sales, building-level rentals and short, recent windows. If the micro facts argue with the suburb-wide picture, trust the facts. Algorithms gather the medians, and they don’t compare apples with apples.
Investors Should Stress-test With Raw Data, Not a Machine
Balance cash flow with key drivers, such as employment, income, demographics, owner-to-rental ratio and infrastructure spending. Then run a pain test using raw data. What if vacancy doubles for a quarter? What if rents drop? If your numbers only work under perfect conditions, they do not work. Know the facts before proceeding.
• Start with the platform’s pick
• Pull true comparables within a tight radius and a recent time window
• Check rental absorption and days on market for the same micro area
• Scan the NSW Planning websites or the local council for infrastructure projects
• Ask the falsification question: what would prove this wrong
• Write the answer down and act on it
Questions and Answers about Algorithms and Property Buying
A good buyer’s agent looks beyond the numbers, assessing local infrastructure, community development, government investment, and emerging trends that data platforms might miss. They’ll also consider the raw data when doing their due diligence.
Are AI property platforms reliable for investors?
Not entirely. They offer starting points, but without human insight and contextual understanding of data, it’s advice can lead to poor decisions or missed opportunities, potentially costing consumers $100,000’s.
How do I avoid buying in a ‘no-go zone’ that actually has potential?
Look for a buyer’s agent who knows the ground. Experience, local connections, and independent research are crucial for identifying true opportunities that others overlook.
Why is human interpretation of data still essential in property investing?
Algorithms can’t see neighbourhood nuances. A human can detect changes in the local market, sentiment, and development plans, all of which matter in a sound property strategy. The raw data is solid; algorithms or AI do not influence it.
Book a Call With One of Our Experts
Paul Mulligan is the Principal Buyers Agent & Expert Negotiator at MULLIGAN Buyers Agents, Sydney. One of only two buyer’s agents in Australia to complete Negotiation Mastery at Harvard Business School, Paul combines advanced negotiation skills with over 30 years of market expertise to achieve outstanding results for property buyers. Passionate about helping clients avoid costly mistakes, he provides expert guidance throughout the buying process from search to settlement.
