Source – theurbandeveloper.com
Runway’s algorithms determine what key property attributes best match customer profile attributes from purchasers. The algorithms then make two predictions:
- Which prospects are more likely than others to purchase a new home, lot, house and land package or apartment?
- What property types (with specific property attributes) are these prospects most inclined to purchase?
Over time, the machine learning engine tests its established matching attributes against purchases, automatically calibrating itself based on the what it has learned, gradually improving the prediction.
“Traditionally, marketers used segmentation to predict consumer behaviour. Machine learning allows us to get down to the individual, determining what is the right product mix for them, and most profitable for the business – in real time.” Director of research and development at Runway, Rowan Kelly said.
Another goal of our machine learning research is to optimise the property offering mix to achieve optimum sales and profitability outcomes. For example, with house and land packaging, the algorithms provide a score against the saleability of a package.
The packages that score poorly can be removed while higher scoring packages can be offered. The machine learning algorithm evolves its scoring automatically over time against ongoing sales, incorporating real world customer buying insights.
Machine learning can also improve productivity within existing processes. For example automatically siting a home plan on irregular lots within our siting software.
So instead of manually-positioning the house, the software will position the house based on what it has learnt from previous sitings. Our goal is for the software to recommend the best possible siting outcome versus a human manually performing the task. This allows the consultant to focus on the customer, rather than fiddling around with software.
