Everyone wants to know how to improve customer experience to increase loyalty, reduce churn and connect with their customers. It’s the question on many retailers’ minds, as they create new positions, like Chief Customer Experience Officer, and start funneling more resources into personalization efforts.
Many retailers realize they need to better understand every aspect of the customer journey to deliver true personalization, but there’s more to knowing how to improve customer experience than social data, surveys and meetings. Beyond static data points, what if retailers could include the same behaviors in-store employees use to understand the customer and tailor their service? What if the time a customer spends lingering by a shoe display or picking up a handbag was part of the equation?
There is a wealth of information about the customer journey hidden in plain sight: customers’ website visits.
Clickstream datasets provide a rich source of information that can capture top-of-the-funnel behavior. Clickstreams digitally echo the footprint of a customer walking around a store—they capture what catches the customer’s eye, what they stop to consider and how long they consider it. Retailers can use this information to understand their customer consideration cycles, the products they interact with and the categories they prefer in a way that offers a more holistic view than traditional transaction data provides.
The importance of time in the consumer’s consideration of products necessitates a machine-learning approach that can model sequential data. Recurrent neural networks (RNNs) are an effective way to model the likelihood of a purchase. This framework can predict if a customer is becoming more or less likely to purchase. In addition, the potential for future enhancements of item recommendation using sequence-to-sequence RNNs offer a multi-faceted approach to customer experience.