An Overview of Conversational AI
Conversational AI is about using messaging apps, speech-based assistants and chatbots to automate communication and create personalized customer experiences that are infinitely scalable.
Over the past few years we’ve seen the rapid adoption of messaging platforms such as WhatsApp and WeChat. Consumers are increasingly using these platforms in place of email, voice calls and face-to-face communication to talk with friends and family. Younger people, in particular, are drawn to using messaging instead of email and voice calls. Not only that, social events are being organized via messaging apps, texting is giving way to messaging, and early adopting consumers are using voice interfaces to make purchases.
Going forward, consumers will increasingly communicate with businesses on their favorite chat platforms, just as they currently do with friends and family. For businesses like yours, that’s critically important. It means that you will need to be able to hold seamless, synchronous conversations with consumers across whatever channel they happen to be using at the time, regardless of whether they’re at home, in their car or on the go.
In fact, according to studies by Twilio and Facebook, messaging-based chat is already emerging as a preferred channel for early adopter consumers to communicate with businesses. And while this is initially happening on messaging platforms such as Facebook Messenger, the shift in how consumers communicate with businesses won’t be limited to text-based chat. With advances in voice platforms such as Amazon Alexa and Google Home, and as automated voice platforms become more widely available, chat will become a channel for business in both text and voice formats.
A Closer Look at Conversational AI
To implement conversational AI, you will need to redesign your customer experiences around natural language dialogue. The nature of that dialogue is that it’s back-and-forth, helping foster an ongoing relationship with the user. It will typically use an array of technologies including messaging, chatbots, speech recognition, natural language processing and artificial intelligence (AI).
The increasing use of AI to power messaging and voice chat will allow you to use conversational AI to create unique customer conversations at scale. However, not every conversational AI interaction will provide full automation, nor does every interaction require it. Instead, many interactions will begin or end with human-to-human live chat. For example, if an automated chatbot is unable to solve a problem, the interaction may be escalated to a human chat operator. That said, over time an increasing percentage of these interactions will become fully automated.
Today, a range of automation is already being applied: from simple rules that can handle a limited number of requests and responses to sophisticated AI-enabled chatbots and assistants that can carry out a variety of tasks or answer a range of questions within a specific domain. Over time, AI will increasingly be used to scale interactions across multiple users and platforms. This will allow companies to leverage stored information about each user from previous conversations and to deliver natural, personalized and gratifying user experiences.
Early examples of companies utilizing chat-based interfaces include in-app notifications for hosts and guests using AirBnB; Facebook Messenger bots; as well as using text messaging to order an Uber, Amazon Alexa to play music or Google Home to control your Nest thermostat.
Why Conversational AI?
Our experience with a range of growth-stage companies indicates that conversational AI applications can lead to:
- Lower customer acquisition costs (CAC)
- Creating engaging experiences in more places
- Better overall user experience
- Higher levels of collective intelligence
In fact, by moving from web interactions to messaging interactions, one company we work with was able to double the amount of time that its customers spend engaged with its brand. At the same time, the company reduced its CAC by servicing more customers without increasing staffing costs.
Not only can a customer have a more engaging experience through chat (and one that potentially leads to higher conversion), businesses using chat can also begin to multiply that experience across their entire user base, as well as multiple platforms, devices and channels. For example, once a business has succeeded at using a chatbot experience to convert one user, it can be repeated for similar users and contexts, potentially even for all one billion Facebook Messenger users.
Another advantage of conversational AI experiences is that they are more engaging than experiences through other channels. Because individual users can be identified through their messaging profiles, interactions can be highly tuned and personalized. Personalization reduces the friction that often occurs when customers have to explain who they are and what products they have each time that they interact with someone new. With a persistent messaging channel that carries context forward, the focus of the interaction (automated or human-driven) can be on addressing the task at hand rather than trying to identify the user or recall account information.
Additionally, participants are able to pause the conversation or restart it at a later time without having to replay the entire process. For the consumer, this eliminates the hassle of dealing with overworked call centers since a message can be sent immediately when an agent becomes available.
Moreover, as the level of automation increases, a growing collective intelligence can improve the conversational interactions that take place. In traditional human-to-human business interactions, customers can only benefit from the experience and knowledge of a single agent. However, when customers interact with a chatbot, they can potentially benefit from every prior interaction that the chatbot has had.
If you want to take advantage conversational AI, start by focusing in on areas of the business that currently require a significant amount of human-to-human interaction. This could include customer-facing activities such as marketing, sales, customer onboarding, product support, or delivering insights or reports. For example, an e-commerce platform that currently provides marketing insights for merchants via a web report, might instead provide those insights to merchants on their mobile device via a chat interface. The company could also enable those merchants to quickly deploy a chatbot that has knowledge of their products and customer shipping status, freeing them up from having to devote resources to answering consumer questions.
Once you have identified the area of your business to focus on, it’s important to prioritize the opportunities based on potential positive impact on the business, as well as estimated difficulty and cost to implement. Examples of the types of impact introducing conversational AI could have include bringing in new customers or new revenue from existing customers, reducing the cost of customer onboarding, or driving higher levels of customer engagement leading to reduced churn and increased customer lifetime value (LTV).
Next, it’s important to determine how conversational interactions can make an impact. Identify the specific metrics that you can improve, such as CAC, revenue or LTV. Once you have done so, focus on the drivers behind them such as engagement (e.g., the length of engagement with a customer), personalization, novelty, friction and automation (to drive down costs).
In addition to understanding the business process to focus on and the metrics for success, it’s important to clarify who the audience is, where they are (i.e., what messaging platforms they are using) and how best to reach them. That understanding will come from mapping out the entire customer journey and identifying the primary sources of friction and frustration for the customer or user as part of that process (e.g., mapping out the steps involved in opening a customer support request ticket). We recommend that you focus on a specific subset of the process first and implement that, such as getting a status update on a customer support ticket, while ensuring that the long-term goal is to cover the end-to-end process.
If you’d like to learn more about some of the key considerations when developing your own strategy for conversational AI, check out our Principles of Conversational AI.