FARMERS USES ANALYTICS TO IDENTIFY HIGH VALUE CUSTOMERS

THE CHALLENGE

In line with its business strategy, Farmers (a New Zealand department store) wanted a way to better predict which of its customers were likely to make a purchase in a high value and high margin category.  The goal was to identify these valuable customers and enable more targeted marketing in an effort to drive sales.  Farmers assumed that the high value purchases were made on a reasonably random basis, so another goal of the project was to determine whether or not there were actually any patterns to capitalise on.

THE SOLUTION

Datamine worked with Farmers to create a predictive model that would be able to forecast which customers are likely to come in store to purchase a high value product.  To build the model, Datamine used a neural network, which is a machine learning technique that enabled the team to pull out patterns held deep within the transactional data.

In doing the analysis, we found that there were patterns in the data that help predict someone’s likelihood to make a high value or high margin purchase – things like the time since their last purchase in other store categories, age, shopping behaviour and other valuable variables.

THE RESULT

Using the model, Datamine successfully identified customers likely to make a high value or high margin purchase.  The goal is now to engage and drive conversion to purchase, as some of these sales may go to a competitor without such targeted intervention.  The neural network making these predictions can be run regularly for ongoing accuracy as customers send different signals via their shopping behaviour over time.  In deploying the model, the team has determined the variables that are most likely to lead to a customer purchasing a high value item, allowing Farmers to market more effectively to specific groups moving forward.