Lightstone is turning three decades of South African property records into AI-driven valuations that estate agents say are reshaping how deals get priced in 2026.
Turning Data into Pricing Power
Data alone is noise, according to Hayley Ivins-Downes, who spoke at the eXp Conference about how raw numbers only become useful when they feed pricing intelligence and buyer insight. With roughly 149 zettabytes of global data now generated annually, the challenge is not collecting more information but filtering it into actionable advantage.
Lightstone’s database covers about 8.56 million residential and non-residential properties across South Africa, stretching back thirty years. That depth gives the company near-complete national coverage, which it claims translates directly into sharper valuations for banks and agents alike.
Only 30% of those properties carry bonds, a figure many assume is far higher. Estates command roughly 50% price premiums over freehold homes, while Gauteng leads in transaction volume and the Western Cape shows the strongest growth momentum. These structural patterns tell agents where demand concentrates and where lifestyle premiums apply.
The Western Cape Premium
Semigration accelerated after 2020 as remote work untethered buyers from urban centers. In the Western Cape, properties now achieve close to 91-92% of asking price, compared to under 90% in Gauteng. Days on market average in the mid-70s to 80s in the Cape, versus around 90 or more in Gauteng.
Demand concentration creates pricing resilience. Agents who understand this dynamic can position stock differently, price more strategically, and manage seller expectations with data rather than emotion.
Over the past decade, the average buyer age has risen from around 30 to 38. That shift signals increased affordability pressure and delayed entry into property ownership. Agents must now engage financially mature, often dual-income, buyers who may be more risk-conscious and data-driven.
AI Models Drive Lending Decisions
Lightstone’s Automated Valuation Model (AVM) powers lending decisions for banks that have relied on the system for two decades. Now AI-enhanced and recalibrated monthly, the model produces two core metrics: a Confidence Score measuring valuation accuracy and a Safety Score indicating the probability that value is not under-estimated.
If a property is valued at R1 million, the system might show 75% confidence within a defined range and 88% safety that value exceeds a threshold. Banks set internal rules around these scores to approve bonds faster and with less friction.
Beyond Deeds Data
Lightstone integrates satellite and drone imagery, roof type detection, pool identification, and area-under-roof measurements often missing from traditional deeds records. Weather and fire risk modeling further enhance valuation precision and insurance assessments.
Transaction volumes dipped during COVID, surged during record-low interest rates from 2020 to 2022, tapered during rate hikes, and are now flickering upward again in 2025-2026. Agents must understand the lag in deeds data and interpret momentum shifts early to stay ahead of cycles.
Days on market and percentage achieved of asking price remain powerful pricing tools. Gauteng’s longer sell times and lower achieved percentages make overpricing dangerous, while Western Cape stock moves faster at higher achieved percentages, supporting strategic pricing that protects margins.
Trust and Transparency
Lightstone points to three foundational pillars: transparency, integrity, and outcomes. The company is Africa’s only member of the European AVM Alliance, a detail it says reinforces its commitment to global standards in automated valuation.
