Over the past months, as call centers across the globe have had to mobilize urgently to respond to a surge of inquiries under the added strain of reduced staff and/or remote working conditions, text based chatbots have been successfully deployed to take up the slack – deployed at scale for multiple use cases for rapid response and recovery across different industry sectors. For example:

• Medical and health insurance companies have built chatbots to provide specific COVID-19 advice.
• Specialist retailers have deployed chatbots to meet an increased demand for fulfilling customer prescriptions online.
• Governments have relied on chatbots to process new unemployment claims.
• In the US, banks have used chatbots to handle a deluge of questions regarding loans and grants mandated by the government’s emergency Coronavirus Aid, Relief, and Economic Security (CARES) Act.
• In the education sector chatbots have helped homebound students get access to the equipment they need to continue their studies online.

The rapid mobilization of chatbots underscores their importance in modern digital communication. A technology once seen as a shiny toy by many executives has come to be acknowledged as an invaluable member of the team.

Yet, with newfound respect and widespread adoption comes greater responsibility. As chatbots graduate from experimental pilot projects to critical communication tools, they must demonstrate not just consistent value to the business but that they can be trusted by their customers.

Based on KPMG data scientists’ experience designing chatbot strategies and implementing them within businesses, we believe that customer trust in chatbots can be achieved by considering and addressing the following four questions.

Pop-up chatbots can have an immediate impact in relieving stress points in the organization but only if the need is scoped properly. In the immediate consequences of the COVID-19 situation, for example, many organizations found their customer service representatives were not well equipped to work from home. They turned to chatbots both to meet existing customer service requirements and to solve new problems brought about by COVID-19.

Start with an analysis of the workflows and human resources that are already in place to gain clarity on what conversation and information gaps a chatbot is needed to fill. Chatbots can only successfully solve customer queries, for example, if they are given access to the right data sources to address common inquiries.

All too often, chatbots are seen as shrink-wrapped solutions when, in reality, they need to learn customer interactions and language, and to adapt quickly to a real-life environment to succeed. Even though chatbot designers can try to anticipate the nature of customer interactions it is important to iterate and improve the programming once the chatbot is communicating live with customers.

For pop-up chatbots to demonstrate value quickly, companies must identify and remedy any immediate roadblocks to adoption.

These barriers could relate to the enterprise infrastructure in which a chatbot needs to be integrated, the data architecture it must communicate with, or the information security issues that arise from its deployment, particularly if they are cloud-based.

Where possible, working within the parameters of existing communication and networking technologies and platforms will help to address likely concerns raised by risk and governance teams. After all, there is no point in fast-tracking a pop-up chatbot solution if it will take months to implement the architecture to support core use cases.

Pop-up chatbots need a formal role in communication workflows to be effective, including a description of what they will do and what they will pass along to human teammates.

Employees – such as those fulfilling front-line customer service roles – need to be comfortable with the support that chatbots offer. They need to understand and be reassured about how a chatbot will complement, rather than replace, their customer support activities.

Sometimes barriers arise if a chatbot solution is seen to be managed by another part of the business, for example by the IT function rather than the customer service function. Leaders therefore need to explain clearly how the deployment of chatbots will help existing employees perform their jobs better by freeing up time to focus on more valuable tasks for customers.

It is not just employees who have to trust their new chatbot teammates. It is vital that customers trust the information being given out by a chatbot. As a minimum threshold, information must be correct and the user experience must be consistent. Just one wrong answer, or a simple mistake such as providing a web link that does not work, can erode customer trust.

Creating a chatbot persona also helps to generate trust. Customers often engage in small talk with bots, perhaps to gauge its intelligence or how perceptive it is. Intelligent chatbot design pays close attention to a chatbot’s tone of voice, demeanor, personality and the language it uses so that it creates an authentic and trusted experience for customers.

The key is conversational consistency. If a chatbot is typically friendly and supportive, but occasionally responds to a customer question in an abrupt manner, or with an answer that seems out of place, then customers begin to lose trust. To support consistency in chatbot design, define and use a chatbot persona description and appoint a person on the development team to review each bot response against it.

The past 3 months have demonstrated that chatbots offer enormous potential for improving customer service and relations even after the COVID-19 situation has passed.

Organizations should consider ‘seizing the moment’ to implement solutions that not only address immediate and critical pain points, but which also help to gain and retain the trust of the customers long into the future.