A key technology used to aid in call deflection within a contact center is the chatbot, a text or voice-based interface that is deployed on the website or within an application to simulate conversation with users and seamlessly support users. Many of these are self-learning bots, and use natural language processing (NLP) and machine learning (ML) technologies, which allow the bot to converse with customers in a conversational tone, providing answers and solutions to a range of customer inquiries, problems, or service issues.
The bots of today can automate tasks, understand words and phrases, frame appropriate responses, and learn from the received inputs, allowing organizations to replace or augment human workers, particularly for common or relatively simple tasks. Some of the key players in the space include Nuance, Amelia, Verint, Kore, Inbenta, Artificial Solutions, and others, along with any number of CX platform providers, which often incorporate AI-based chatbot functionality. For companies seeking to build their own chatbots, companies such as Google Dialogflow, Amazon Lex, IBM Watson Assistant, Facebook’s Wit.ai, and Microsoft Azure Bot Services provide the tools and platforms to create customized virtual assistants.
Because AI-based bots can understand language and intent, they have significantly more functionality and polish than traditional, rule-based bots. These bots can perform various tasks, such as conducting sentiment analysis, predicting consumer likes and dislikes, assisting the customer with the right product or service, and, if a stumbling block occurs, can quickly route the inquiry to a live human for resolution.
Not all chatbots are created equal, however, so several assessments should be made to ensure that the bot technology that is being purchased, modified, or built from scratch internally can meet the following attributes:
- Responsiveness: An artificial intelligence (AI) conversational bot should be able to reply fast as soon as it receives inputs from the user, which portends the use of streamlined backend integrations with various data sources and applications, to reduce or eliminate the time it takes for a bot to respond to a customer’s input.
- Response accuracy: Responsiveness alone is worthless if the accuracy of the responses or answers that bots provide is low, unclear, or incomplete to the end user. Further, the bot must be tested to ensure that not only are the appropriate responses to queries returned, but that the bot is properly tuned to ignore inappropriate, insensitive, or objectionable inputs that may elicit unwanted responses.
- Clarity and error handling: A bot’s ability to deal with the errors and its ability to recover from those errors must be tested, so that if a bot fails to understand user inputs, it can ask the customer alternative questions for clarification, or should immediately connect the user with a live agent.
- Tone and personality: A bot’s voice (whether deployed via voice or text) should be relatable and fit within the character and tone of the user base, so that users feel comfortable conversing in a normal manner. A bot deployed within a financial services use case is likely going to differ in tone and feel from a bot deployed via a streaming music service.
- Ease of navigation: The navigation flow of a bot should be tested to ensure the customer does not feel lost while speaking with the chatbot, and can easily access the information without navigating through decision tree-like menus.
- Recall and intelligence: Abot’s intelligence must be tested to ensure it can recall the information presented to it by the customer, and then provide the correct response, using a combination of knowledge bases, supervised ML, and a single source-of-truth for important policies or procedures.
- Understanding: A bot should be able to understand all requests, small talk, vernacular, idioms, and emojis sent by the user to frame appropriate responses.
- Omnichannel and device compatibility: A bot must be able to perform seamlessly across all devices, platforms, and OS versions, and communication channels, with a similar experience and feel. Customers should be able to seamlessly switch to another bot or a live human without needing to re-identify themselves, or repeat information.
- Multithread understanding and execution: Customers often have more than one task to accomplish, and bots should be able to identify multi-process and non-related separate queries, and seamlessly be able to handle them in a logical manner. This is essential to creating bots that can more efficiently handle more complex inquiries and further improve live call deflection rates.
- Security: Data security is a major concern for all enterprises, and bots should be able to be integrated within the organization’s security framework to ensure the data being provided to the bot is properly handled and secured, particularly when it is regulated information (such as healthcare or financial data). Organizations should deploy frequent security testing to find and eliminate security loopholes or vulnerability issues.
Author Information
Keith has over 25 years of experience in research, marketing, and consulting-based fields.
He has authored in-depth reports and market forecast studies covering artificial intelligence, biometrics, data analytics, robotics, high performance computing, and quantum computing, with a specific focus on the use of these technologies within large enterprise organizations and SMBs. He has also established strong working relationships with the international technology vendor community and is a frequent speaker at industry conferences and events.
In his career as a financial and technology journalist he has written for national and trade publications, including BusinessWeek, CNBC.com, Investment Dealers’ Digest, The Red Herring, The Communications of the ACM, and Mobile Computing & Communications, among others.
He is a member of the Association of Independent Information Professionals (AIIP).
Keith holds dual Bachelor of Arts degrees in Magazine Journalism and Sociology from Syracuse University.
