From Responding to Resolving: The Evolution of AI and Customer Service

Touted as a way to improve customer service, help-desk AI agents have been the industry’s shiny new object for a few years. And while they have been created with the right intention of providing immediate responses to customer questions and issues, and also reducing the resources of strained call centers, most have been underwhelming at best and more often, leave the consumer more frustrated than they would have been waiting on hold.
Most commonly, AI agents have failed to understand what a person needs or has not had the authority to take action, and simply pinch hits to a human (who has often had a repetitive exchange with the customer). But just because they haven’t met expectations yet, doesn’t mean it’s time to walk away. As the saying goes….Rome wasn’t built in a day, right?
AI holds so much promise to completely transform customer support, and when done correctly, will enable consumers to get issues resolved quickly, on their terms, 24/7. So what will make AI powered automation work?
  • Deep Learning: First and foremost, AI needs to learn from real exchanges, like historic emails and call center logs (in addition to FAQs and other explicitly trained scenarios). This will ensure the AI is able to understand the range of which people ask about a single issue or topic, and how an agent has responded in the past. The proper response can even be determined based on various contextual factors, like a person’s loyalty level or size of wallet, to provide the most appropriate response every time.  Deep learning also enables AI to continuously learn and improve over time.
  • Authority to Resolve:  AIs need to be able to resolve issues. The first step is to accurately understand what a person needs (see point above), the second step is to close out tickets doing things like issuing a return, rebooking a missed flight or upgrading a seat or hotel room. Certain scenarios are more sensitive than others (i.e. a complaint about an issue with another passenger on a flight) and therefore require human intervention, so AI should be used to automate the closure of the low-risk repeatable tickets leaving high-touch queries to humans.
  • Empowered Human Agents: When a conversation is elevated to a human agent, the AI must pass along the specific information that will help the agent provide fast, accurate support. This involves both picking out the specific content from the AI and customer exchange, as well as the information from a CRM, OMS or other system relevant to this exact need. This will ensure that the human agent does not re-ask questions (causing frustration and increasing the resolution time).
  • Proactive Support:  AI can also help companies move from reactive to proactive support. For instance, the AI would notice a delay in the inventory management system and automate a notification to the customers who are currently awaiting the item. This will enable a company to reach the customers before they have to send an email or make a call themselves. Without AI, this would be entirely too expensive and complex to handle with human agents.
AI will usher in a new era of customer support that is proactive, personal and accurate. With research suggesting that the customer experience will become more important than price and product by 2020, and there are currently over 1B customer service tickets created daily, companies must leverage AI for customer support in order to provide the experience that customers expect. An AI Agent can’t just respond, though, it must offer a high quality resolution to deliver on the true promise of AI.