Automated valuation models (AVMs) powered by artificial intelligence may be costing UK sellers tens of thousands of pounds, with estate agents warning the tools routinely undervalue homes, especially in rural, Northern, and lower-income areas.

A new poll of 250 estate agents by agency software platform Alto found that 87% believe AI-driven valuation tools regularly fail to reflect a property’s true market value.

Yet despite the widespread mistrust, 78% said they use AVMs when pricing homes, with nearly one in four (23%) relying on them exclusively.

Nearly three quarters (73%) of agents admitted they don’t fully trust the tools they’re using, and almost a third (28%) said they typically adjust AI-generated valuations by £10,000–£20,000. One in ten agents reported corrections of over £20,000, pointing to significant potential losses for homeowners whose properties are priced solely using automated systems.

UNDERVALUATION RISKS

One in five agents said rural properties are frequently undervalued by AI, while 11% cited systemic undervaluation in Northern and lower-income areas.

Estate agents in the North East, East of England, and Yorkshire and the Humber were the most likely to report significant discrepancies between AI estimates and actual market value.

In the North East, 11% of respondents said they regularly adjusted AVM pricing upwards by £20,000 or more to better reflect local market conditions.

A spokesperson for Alto said: “There’s a real risk that properties are being priced using flawed or generic data. Agents are telling us that sellers could be losing out – and we should be listening. AVMs are no substitute for boots-on-the-ground expertise.”

KEY FACTORS OVERLOOKED

Nearly eight in 10 agents (77%) believe AVMs often overlook key contributors to property value, particularly recent improvements like extensions (26%), loft conversions (26%), and smart home technology (25%). Other overlooked aspects include noise levels, crime rates, and local economic conditions.

Agents were particularly critical of AVM performance on homes in rapidly changing markets (27%), rural settings (23%), and transitional neighbourhoods (21%), as well as unique or non-standard properties (32%).

Despite frustrations with their accuracy, two-thirds of agents said they continue to use AVMs primarily to save time. However, 37% acknowledged that the industry may be over-reliant on these tools.

Alto’s spokesperson added: “AI should support professionals, not replace them. Selling a home is often a deeply personal and emotional experience. Ensuring sellers receive fair value means retaining the human insight and local knowledge that only an experienced agent can provide.”