Every customer service interaction can decide a customer’s loyalty. According to a study by Zendesk, over 50% of customers have switched to competitors after just one bad experience. In a guide published on its blog, the company ElevenLabs details the skills that enable agents, both human and artificial intelligence (AI), to resolve problems and build loyalty.
These skills fall into two families. ‘Soft skills,’ or interpersonal skills, include communication, patience, empathy, and adaptability. Technical skills encompass product knowledge and incident management. The guide details seven: communication, problem-solving, conflict management, empathy, adaptability, personalization, and product knowledge.
Humans and AI Agents Facing the Customer
For a human agent, communication involves listening and word choice, avoiding jargon. The example cited in the guide illustrates this approach: ‘Let me make sure I understand: your order shipped on Monday but never arrived.’ A voice agent based on a large language model (LLM, a system trained to manipulate language) can hold a fluid conversation, but it needs precise rules to manage turn-taking and interruptions.
Problem-solving follows a comparable logic. A human agent diagnoses the situation, checks the details, and proposes a solution. An AI agent, on the other hand, relies on available data, evaluates possible leads, and selects the one with the highest confidence score before detailing the steps to follow. In the guide’s example, a declined card payment is checked by a human advisor, while an AI agent connected to account data identifies the most likely cause.
When tension rises, conflict management becomes decisive. The guide emphasizes that AI can spot frustration through natural language understanding (NLU) and transfer the conversation to a human, along with the chat history. Regarding empathy, the text specifies that AI does not feel emotions, but it can identify signs of frustration in voice or text and adopt an empathetic tone. Jannes Koopmann, Sr. AI Business Owner at 3Shape, stressed the importance of an empathetic tone and a human feel in modernizing his company’s support.
Improving Through Practice and Feedback
According to the guide, these skills can be learned and measured. For human teams, this involves tracking customer satisfaction (CSAT), first-contact resolution rate, and average handling time, then working on weaknesses through role-playing and call observation. For AI agents, the guide cites the containment rate, i.e., the share of requests handled without human intervention, as well as the resolution rate.
Before any deployment, changes must be tested in a simulation environment, including edge cases, particularly those requiring a transfer to a human after a tone shift mid-conversation. ElevenLabs indicates using a testing framework that turns shortcomings identified in existing conversations into evaluation cases.
Finally, the guide reiterates the division of roles. AI agents excel at applying policies and can automate technical tasks like consulting customer files or processing refunds. Humans retain final judgment on aspects involving interpersonal skills. A well-conducted transfer, the text specifies, passes on the conversation context so the customer doesn’t have to repeat everything.
