Farfetch Chat R&D Overview
High-fashion marketplaces require top-class customer interaction. Users of such online platforms demand a reliable, precise and timely service, but also expect the best in class customer experience throughout the customer journey. Therefore, a frictionless experience with a high-touch feeling is key for clientele satisfaction. As our high-end fashion business continues to grow, our services are more than ever under increasing pressure. More recently, with the impact of COVID, it created an additional demand that was not anticipated.
So it is without surprise that scaling up FARFETCH business to our ambition while maintaining our customer’s loyalty is a challenge. Recent findings show that live, human, conversation services (i.e., WeChat) - the main shopping channel in China at FARFETCH - translates into significant conversion rates (CR) improvements.
We are currently witnessing an enormous paradigm shift from the traditional search and click search thanks to conversational assistants in online shopping and other social media-related tasks, especially in the younger generations. WeChat, the largest social network in China, and WhatsApp, now a core service in Facebook, are the most clear signals of this paradigm shift by closing the gap between consumers and brands.
The challenge addressed by Farfetch Chat is to mimic a fashion specialist that understands the customer needs and provides fashion advice leveraging the vast textual and visual data together with knowledge accumulated by past experiences with a massive number of users. Our vision is to make a step-change in the online high-fashion marketplace by advancing conversational AI technology with multimodal capabilities, see Figure below of a mockup.
Figure: A mockup version of FARFETCH Chat.
For an e-commerce marketplace, task-oriented multimodal conversational agents (MCA) have the potential to make a groundbreaking shift in the way users do their online shopping.
A key element of the project is the deployment and validation of a solid demonstrator. To achieve this outcome, we will run validations with focus groups at given instants. We will quantify (1) the customer perception of FARFETCH Chat value and (2) collect statistics concerning the actual changes in customer behavior through several indicators, such as % of products viewed and added to the shopping cart, and customer survey for quality control.