3 Ways Conversational AI is Transforming the Customer Journey
Siri, Alexa, and Google Assistant have changed the customer experience forever. As consumers use these smart devices every day, they expect similar human-like and effortless automated conversations when interacting with companies. Hence, many contact centers are quickly adopting Conversational AI, to revamp the customer experience and automate routine tasks.
What is Conversational AI?
Conversational AI is a set of technologies, including machine learning (ML), natural language processing (NLP), automated speech recognition (ASR), and text-to-speech (TTS), that enables human-like, natural interaction between people and machines. It allows machines to understand human speech and text, predict intents and sentiments, and interact naturally in various languages.
Expanding on the components of conversational AI, Natural Language Processing (NLP) uses machine learning to examine patterns in natural language data, using these patterns to improve a computer’s language comprehension. Machine Learning creates systems that can learn and improve using feedback loops and without requiring additional software code development. Some examples of NLP platforms are Google Dialogflow, IBM Watson, and Microsoft Cortana.
Contact centers are ideal for implementing conversational AI applications. Why? Because the volume of interaction data they generate daily is ideal to develop, test, and tune machine learning models. Training AI models is key for gaining efficiencies, which in turn helps optimize the customer experience.
Conversational AI solutions like chatbots have been around for a few years now. More recently, with more contact centers going to the cloud, newer applications like voice-bots, real-time agent assistance, automated call scoring, NLP-based call steering, and sentiment analysis are becoming more mainstream. Now, let’s look at the three most common use cases for Conversational AI in the contact center.
Today, customers are looking for faster resolutions and want to self-serve. According to a report from Harvard Business Review, 81% of all customers attempt to take care of matters themselves before reaching out to a live representative. Conversational AI solutions like Five9 Intelligent Virtual Agent (IVA) can classify conversations based on customer intents and assist by providing automated responses. IVAs can easily answer routine questions (When will I get my refund? What is my order status?) using natural language responses over voice or digital channels, seamlessly escalating to a live agent (if needed).
IVAs can also deflect calls from going to an agent by automating tasks without any human intervention. For example, if a customer tells the IVA, “I want to change my delivery address,” the IVA can automatically perform that update in the CRM. It can also support advanced use cases involving multi-turn conversations such as booking a ride or ordering a pizza with toppings. Check out several case studies that validate the effectiveness of IVAs for complex intents.
Even if the goal is to replace an archaic IVR system to provide a natural way for customers to express their inquiries, an IVA solution uses conversational AI to route the call based on the user’s tone, biometrics, and other contextual information.
During a Live Call with an Agent
Conversational AI-enabled solutions such as Five9 Agent Assist empower agents during a live interaction by providing real-time transcriptions and smart contextual guidance to provide better customer experience. Agent assist solutions also automate tasks and summarize the call, which enables agents to wrap up the after-call-work sooner. TruConnect was an early adopter of Five9 Agent Assist and using automated summaries and live-guidance features, they saved up to 30 secs/call in AHT (Average Handle Time).
The AI engine understands the conversation between the agent and customer and triggers workflows (like looking up case notes, call history, order backlog, etc.), or it can even remind the agent to perform specific actions (reading compliance statements, escalations, HIPAA verification etc.).
When an agent is well informed and efficient during the call, they provide the best experience for your customers while reducing handle times, wait time etc. The key difference between Agent Assist and IVA is that the conversation is always happening between humans in Agent Assist and the bot is assisting the agent. In contrast, with IVAs, the bot is assisting the customer.
After an interaction
Conversational AI can be used to understand the quality of every conversation by producing a call score metric. Call scoring is not new, but when a contact center handles thousands of calls every day, it is impossible to listen to every call and produce a scorecard manually. Instead, AI can understand sentiments based on spoken words and use tonal analysis to include meaningful metrics that can be used to evaluate an agent’s performance and provide accurate feedback. In some cases, conversational AI can be used to understand human speech patterns to isolate conversations that yielded positive results in the past (like closing a sale, higher CSAT scores, etc.) and use that information to guide an agent during the call to produce a similar positive result. Solutions like Five9 Interaction Analytics can turn customer conversations into actionable insights for better customer engagement.
Five9 builds conversational experiences by harnessing the power of the most advanced AI technologies, including Google Dialogflow, Amazon Lex, IBM Watson, and more. For example, Agent Assist is powered by Google CCAI APIs for speech detection and NLP (Dialogflow). Google is leading this space with their low WERs (Word Error Rates) and highly accurate natural language intent detection using Dialogflow, which supports a broad set of languages besides English. With Five9, you get the best of breed technology in AI without worrying about the complexities of setting up and running these projects on your own.
Five9 simplifies the deployment of AI and automation solutions using a Graphical User Interface-based approach to unify contact center data with ML models. It enables non-technical users to deploy engagement workflows in hours rather than days or months.
I am leading an initiative at Five9 called the Conversation Architect, and in the next blog post, I will cover what challenges we are looking to address with this technology. Conversation Architect will enable rapid deployment of conversational AI models like Agent Assist, IVAs, and others. It simplifies the underlying complexities with deploying machine learning models and allows non-technical users (CC managers/supervisors) to build, deploy and measure AI performance post deployment rapidly.
Read more about this in the next blog post!