Evo, a ski and sporting goods retailer, is set to launch a customer service chatbot similar to OpenAI’s ChatGPT. The introduction of ChatGPT in 2023 has revolutionized the potential for AI in business, according to Nathan Decker, director of ecommerce at Evo. ChatGPT utilizes generative artificial intelligence and natural language processing to provide accurate responses and content creation.
Decker is impressed with ChatGPT’s ability to understand various languages and the wide range of scenarios it can handle. Unlike traditional customer service chatbots that rely on a fixed set of questions and answers, ChatGPT can interpret and comprehend nuanced and complex language. This flexibility makes it more suitable for retailers like Evo, which require a chatbot capable of grasping the intricacies of customer inquiries.
However, while a generative AI bot like ChatGPT can answer almost any question, retailers must ensure that the bot is designed not to provide unrelated information or redirect queries to appropriate departments. The responsibility lies with the retailer to create sound business logic within the bot.
To develop their customer service chatbot, Evo enlisted the help of artificial intelligence vendor AlBee. The company provided AlBee with training materials and additional brand information to train the bot. The development cost is expected to be under $1 million, with an annual fee comparable to other vendor costs.
Implementing a chatbot to handle customer service inquiries will greatly benefit Evo, particularly during peak winter months when there is a surge in traffic and sales. Evo experiences 40% of its sales in November and December, and the ski season continues until March, resulting in a constant influx of customer service inquiries beyond the holiday season.
While Evo will still hire seasonal customer service employees, the chatbot will handle less complex queries such as return policies or store hours. The company aims to allow its human agents to focus on more complex questions and provide personalized assistance to customers.
Initially, Evo plans to soft-launch the chatbot on pages with high bounce rates during the slow season. This allows them to test the bot’s reactions and responses to real customer inquiries. Once the bot is refined, it will be rolled out to high-traffic areas such as product detail pages and search results.
Evo intends to measure the success of the chatbot quantitatively and qualitatively. Through A/B testing, the company will compare engagement rates, conversion rates, and session value between customers who interacted with the chatbot and those who did not. Qualitative evaluations will involve monitoring interactions, gathering feedback from customers and staff to assess the effectiveness of the chatbot in providing a positive customer experience.
The goal for Evo is to eventually reduce the need for seasonal customer service agents by leveraging the capabilities of the chatbot. While Decker acknowledges that the chatbot cannot replace the knowledge and expertise of human agents, he believes it can significantly reduce the workforce required.
Results from a recent survey indicate that human customer service agents remain vital in driving customer conversions. However, with efficient agents, a chatbot can potentially decrease the workforce by 25%-30%.
Ultimately, Evo hopes that the introduction of their customer service chatbot will improve customer support and streamline inquiries, ensuring a smooth and satisfactory experience for shoppers.
