One of the UK’s leading property data firms says estate agents need only prospect among 10% of their local stock to reach 62% of those most likely to instruct now that artificial intelligence or AI has become more sophisticated.

Milton Keynes-based data firm TwentyEA says AI-based prospecting means agent prospecting is increasingly becoming more accurate and less of a ‘scatter-gun’ approach – although the firm admits take-among agents is still at an early stage.

“The combination of AI and machine learning can provide powerful insights for estate agents, particularly in the area of instructions, by predicting which off-market properties in their area will list in the next four months,” says the firm’s executive director Katy Billany (pictured).

“The beauty of machine learning is that it learns from historic data to forecast consumer behaviour.

“Even a year ago, targeting the top 30% of properties to reach 80% of new instructions was impressive, but machine learning has improved this even further, so that estate agents only need to communicate with 10% of the properties in their patch, to reach 62% of the vendors who will instruct.

“This is a remarkable five times better than random targeting. In fact, I know that some agents are generating over 60% of their valuations from Forecast data.”

Services like TwentyEA’s Forecast tool can tell agents accurately how long the owner has lived in a property; how old they are; whether the property has been withdrawn from the market in the past; or whether a sale has fallen through, but the property has not gone back up for sale.

“This data is then combined with hundreds of other data points that might increase their likelihood to move – are they a young couple with a baby living in a small flat, or someone who is recently divorced living alone in a large house with lots of equity,” says Billany.

Read more: When prospecting goes wrong.