When Lightstone took South Africa’s first Automated Valuation Model (AVM) live with Absa in 2006, most banks assumed that automated valuations would be useful for low-value, low-risk decisions, but never as a serious substitute for a human valuer walking the property.
Twenty years later, AVMs sit at the centre of mortgage origination, portfolio monitoring and capital management at every major South African lender. That is the story most people would tell about the last two decades. I think it’s the wrong story.
A market once shaped by scarcity of information is defined in 2026 by an abundance of data. Information asymmetry has collapsed, decision-making has become smarter, faster and more defensible. Property intelligence influences more than transactions alone. The real shift, then, hasn’t been the rise of automation but a recalibration of trust.
Who knew what?
In 2006, buyers generally relied on instinct and what Estate Agents told them, while sellers often priced based on anecdote rather than evidence. Estate Agents were the gatekeepers of market intelligence because they controlled the local comparables, relationships and market context.
Financial institutions operated differently then: lending decisions relied on manual, site-based or drive-by valuations, and consistency varied across valuers and portfolios. Two near-identical properties could carry materially different values simply because they were assessed by different people on different days. Banks were valuing properties one at a time. Agents were working off paper deeds searches. Buyers and sellers had no objective reference point.
The market worked, but with friction, delay and asymmetry. Lightstone recognised that if information could be cleaned, modelled and delivered effectively, it could change how the market made decisions.
Each gatekeeper earned a premium for what they knew. But the information they guarded, and the premium they earned, have now largely disappeared. A first-time buyer in Edenvale can pull a suburb report, a price history, a school catchment and a crime profile before lunch. A seller can get an automated valuation in under a minute. The information advantage disappeared. The balance of influence changed with it.
What banks actually bought
When banks first adopted AVMs, the attraction was efficiency: faster turnaround, lower cost per valuation, and fewer site visits. Those benefits arrived, but what banks actually needed was consistency.
A bank’s mortgage book is only as defensible as the methodology behind every individual valuation within it. Twenty years ago, a portfolio of fifty thousand homes was fifty thousand idiosyncratic opinions stitched together. Today, that same portfolio can be revalued continuously against a single statistical framework. Capital can be allocated against actual risk rather than conservative assumptions. That is not a process improvement. It is a different operating model.
The legislative environment reshaped the market materially. The National Credit Act made affordability central to lending, which increased demand for better valuation data. FICA and later POPIA raised the bar on compliance, while the Property Practitioners Act modernised the regulatory framework for the agency side of the market. Each change increased both the complexity and the value of reliable property data and raised the cost of inconsistent or unexplainable decisions.
Lightstone has had a front-row seat to this transformation. Today the business supports every major South African lender, provides property data and indices to the South African Reserve Bank, and remains the only African and non-European associate member of the European AVM Alliance.
The next decade of regulation will raise that cost further, particularly around model interpretability – the requirement that automated decisions must be explainable and auditable. Climate risk disclosure will add another layer, and pressure will continue to favour institutions whose decision infrastructure is statistical, governed and auditable rather than artisanal.
The value of Estate Agents
Some say that data has made the Estate Agent’s life harder, but I would argue differently. It has made the good ones more valuable and the average ones replaceable.
When comparables became universally accessible, an Estate Agent’s value shifted to judgement: reading the buyer in the room, structuring an offer, knowing when a price needs to drop and when to hold. The Estate Agents who have embraced this change are the ones who have survived and flourished.
The next twenty years – and what data can’t fix
Most forecasts about the next twenty years focus on what AI will bring: sharper valuations, faster underwriting and smarter risk models.
All of that will happen.
But the real challenge is not about data.
It is the affordable housing gap, the title deeds backlog and household incomes that have not kept pace with property prices.
AI will let us value, finance and insure homes with extraordinary precision. It will not, on its own, create more buyers who can afford them. That is a policy, infrastructure and economic problem, and data’s contribution to solving it is more modest than the industry sometimes pretends. What data can do – and what we should be honest about – is make the gaps visible enough that capital, policy and development can be pointed at them with real conviction.
The next competitive battleground will not be access to data itself. Data is increasingly available. The real differentiation will come from how effectively organisations embed that intelligence into workflows, decisions and customer experiences.
A prediction worth making is that the property transaction itself will finally become continuous rather than episodic. Listing, valuing, financing, conveyancing and registration stop being separate events and become a connected lifecycle, with property data behaving like a live utility.
The more things change…
People will still fall in love with homes. Banks will still manage risk cautiously. Estate Agents will still trade on trust.
What has changed – and continues to change – is the quality of intelligence driving those decisions, the speed at which it moves and where trust sits within the ecosystem.
The institutions that succeed over the next twenty years will not necessarily be those with the most data.
They will be the ones most trusted to turn that data into decisions.
- Sara Winstone, Lightstone Property Managing Director
