New build has a significant part to play in addressing the UK’s housing needs, but for lenders, these properties can present a unique challenge when it comes to valuation.
Figures from the Ministry of Housing, Communities and Local Government (MHCLG) show 35,910 new build starts in England between April and June 2026. This is up 6% on the previous quarter and 20% on a year earlier. As a result, a growing number of developments are entering the early phases where comparable data is limited.
Automated Valuation Models (AVMs) have become an increasingly central part of the mortgage process, helping lenders make faster decisions and reduce reliance on physical valuations. Yet the effectiveness of any automated model ultimately depends on the data available and crucially, how that data is interpreted. And this is where new build presents a different proposition. Traditional AVMs typically draw heavily on historic transaction data and comparable sales. For established properties and neighbourhoods, there may be a substantial history of transactions from which a model can learn. A new development does not provide that same depth of evidence.
The new build data challenge
Developments progress through multiple phases, individual plots can have different characteristics. Before a scheme is well established, there may be relatively few completed transactions to use as direct comparables.With more schemes now under way, this is a challenge lenders will meet more often.
The mistake is assuming that because an AVM works effectively for established housing stock, the same methodology can simply be extended to new build. It requires a valuation approach designed around the particular characteristics of the sector, looking beyond historic transactions to create a richer picture of the property, development and surrounding market.
At CLSQ, our approach combines proprietary new build plot-level intelligence and our Desirability Index with geospatial analysis and structured market data. This provides a valuation framework specifically designed for properties where conventional comparable evidence is scarce.
A new build property can’t simply be viewed as another address within an existing postcode. The address is yet to be allocated, and the postcode is to be subdivided. Understanding the development, its progress and individual plot characteristics provides valuable additional context when assessing value.
Filling the gaps intelligently
A specialist model also needs to deal effectively with incomplete information.
There will inevitably be occasions where direct property-level data is unavailable. The challenge is not to pretend those gaps don’t exist, but to have a robust methodology for dealing with them.
CLSQ's New Build AVM uses weighted market averages and confidence-based logic to backfill property attributes where direct data is not yet available. This allows the model to generate a plausible valuation whilst recognising the limitations of the underlying evidence.
This is particularly relevant where developments are evolving and properties are being completed and sold in phases. Geospatial analysis provides further context by examining surrounding market activity alongside development and plot-level intelligence, creating a broader evidence base from which to assess an individual property.
Confidence matters as much as speed
The government’s ‘Your First Home’ scheme, due to be set out during the Budget, would support first-time buyers into new-build homes with a deposit of just 2.5%. With the full parameters set to follow later this month, lenders could see more new-build applications at higher loan-to-values.
The mortgage market is steadily becoming more automated. Lenders understandably want to reduce friction, improve turnaround times and give borrowers greater certainty earlier in the application process.
But the measure of successful automation cannot simply be how quickly a decision is produced. It also has to be judged on how much confidence a lender can place in that decision.
Better property data enables lenders to make more informed decisions about when an automated valuation is appropriate and when further assessment may be required.
For new build, that means combining transaction evidence alongside development intelligence, plot-level attributes, geospatial information and structured approaches for dealing with missing data. It’s a fundamentally different approach from simply searching for the nearest historic comparables.
Building automation around the property
There will always be circumstances where specialist human expertise is required. The opportunity presented by financial grade data and modelling is to identify more accurately where automated decisioning can be used with confidence and where further assessment is appropriate.
A shortage of conventional comparable evidence does not necessarily mean a new build property cannot be valued effectively using automated techniques. It means the model needs access to the right information and must have been designed specifically to understand that market.
This is a key principle as the industry handles a greater volume of new build and more lenders consider where automated valuation can help. We shouldn’t be trying to make properties fit the model; we should be building models capable of understanding the properties lenders are being asked to finance.
New build, with a growing pipeline of schemes now under way, demonstrates exactly why this distinction is important. Getting it right gives lenders a richer evidence base and, crucially, greater confidence in the valuation decisions that follow, and ultimately supports the buyers who rely on those decisions.
