Customer service has come a long way. From the counter at the back of the store to off-shore call centers, to automatic attendants. We are now seeing the merger of artificial intelligence, mobile devices, and smart speakers as the next generation. In general, as technology has improved, so has the speed and accuracy of customer engagement. However, even with all of the improvements, US companies lost $75 billion in revenue in 2018 because of unpleasant customer experiences.

What can organizations with a high level of customer interaction do to further improve this critical part of doing business without digging deeper into margins? For many, the next step in their customer service journey is the deployment of voice bots.

Voice bots are a type of voice user interface, or VUI, that enable customers to access information, receive answers to questions, or carry out transactions by interacting audibly with an automated system. Tools that make designing and running these systems easier are delivering technology to replace or supplement interfacing with humans or graphical user interfaces like websites or mobile apps.


A Voice bot is much more than a voice recognition

To understand how voice bots work and the technology that makes it possible, a good place to start is to look at the standard architecture:

When a user interacts with the application, what they say first goes to a speech recognizer to convert the words into text. The resulting text is then passed to a natural language understanding (NLU) unit that does more than look for keywords or specific commands. It uses various artificial intelligence methods to understand what the user is trying to convey. The deduced information, keywords, and commands are then sent to the dialog manager, which tracks the history of the session and manages the flow of the conversation.

The next subsystem, the task manager, is where all of the heavy lifting is done. Often this is interfacing with other business systems to process a transaction, extract information, or try and solve a problem. Or it may follow a recipe defined for the requests customers commonly make. Once the task manager determines the next step, it needs to communicate with the customer by telling the dialog manager what needs to be done. The return message requirements are sent to the output generator where a text response is created that is as natural as possible. The text is then passed to the output renderer, which converts the message into speech to be played to the customer.

The application of AI, and specifically machine learning in the NLU and task manager, is where the most significant changes to technology are happening. Another area is better integration of other business systems to provide the task manager with context and historical information beyond simple information retrieval.


A better way to automate the customer experience

Companies will be switching more and more to voice bots because text chatbots, applications, and websites require manual input of information. When using audio, the customer does not have to look at a screen or use their hands. 

A recent study by Salesforce showed that 69% of consumers prefer bots for quick communication with brands. The three main reasons given were that they had the ability to interact 24×7, they don’t have to wait for a person, and they can get answers to simple questions quickly.


This most obvious advantage to a business deploying VUI is the ability to manage transactions and answer common questions at a lower cost. Also, well-deployed ChatBots can increase ecommerce sales by upselling and downselling, encouraging users to return to abandoned carts, and feeding more data into predictive analytics and personalization.


Best practices of implementing voice bots

With such significant advantages, most companies with high customer interaction are looking to implement some sort of VUI. Implementations of text-based bots, virtual assistants, previous customer interaction automation initiatives, and early voice bot projects can be used to develop a set of best practices and things you should avoid.

A great place to find out what not to do is our article on “5 Voice User Interface Design Mistakes & How to Avoid Them.”

These are great suggestions for every implementation, but what about specific best practices for different applications? The possible uses of voice bots go beyond the Amazon Echo and other voice assistants—many different companies and organizations can benefit.

For example, if you are providing a way for customers to deal with problems like resolving a missed payment, you need to understand that they are coming to your system in an emotional state. Your solution should be designed with this in mind, including the use of natural language processing to deploy sentiment analysis. This is a tool that lets you discover what the consumer is feeling about the process or the choices presented. Based on this information, you can change their journey in a way that lessens negative emotions.

When looking at systems that automate ecommerce, one less obvious lesson learned is the use of small asks to gain buy-in that gently pushes the consumer through the process of making a purchase. You need to start with simple questions like color or size and work your way towards their personal information and the actual purchase. Never ask for too much information at one time and build on their ever-growing commitment to the transaction.

As a final example of an application, suppose you need to gather a significant amount of information from a user when they are reporting some sort of incident. Maybe you are an insurance company and the customer is reporting an auto accident. A significant lesson learned in previous systems was that users hate repeating information that you should already have. Connect your VUI to your existing database and ask simple questions that will allow you to leverage the existing data. For the insurance example, the policy number for the vehicle involved should be all you need to get personal information about the policyholder. Now you only need to verify that it is still correct.


Give your customers a better experience with voice bots

It turns out that 40% of the questions that customers ask during a customer service call can be resolved by a bot. If your company has a high level of customer interaction, that statistic alone should be enough to start you on the road towards implementing this next generation of customer service automation. The time is now to reduce costs, increase accuracy, and learn more about your customers with a well-engineered implementation of voice bots across your company.

And Botmock is here to help you understand how to deploy your system and to provide you with the tools you need to design and maintain your voice user interface. Contact us today for more information or sign up for a free demo of our platform.


Teams of all sizes are using Botmock