In today’s age of instant expected gratification for customers, the ability to serve users more quickly and solve their problem the first time is highly valuable. So automating some of your support functions has obvious benefits. It allows you to improve quality and resolution times by standardising support processes and making them available 24/7. This, in turn, translates to better end-user experiences and overall increased client satisfaction. Automation also allows you to explore further enhancements to your support function. By analysing and progressively automating high volume but low complexity tasks, your smarter human teams can focus on the more complex, high-value issues.
When it comes to automating the way you support your customers, chatbots have some unique advantages. Chatbots are most commonly used to act as a ‘front line’ for customer queries – responding to FAQs with pre-programmed information and/or providing functionality for a customer to self-serve rather than speak to a human agent.
Why does a chatbot work so well? Firstly, it’s what people are used to. Chabots fit into people’s ‘mental model’ of a conversation and how we solve problems. They support very natural and open-ended interactions – meaning people don’t have to have an intricate knowledge of the issue they’re faced with in order to solve it. Teaching the bots enough natural language techniques means they can deal with problem descriptions coming from web developers and OAPs alike.
From a more technical perspective, they can provide support to users across a range of devices and platforms, from web-based to Facebook Messenger, Slack, SMS or even email.
Sounds good, right? So how do I make sure this happens?
Agility is Key
Moving a great deal of decisions that were typically made by humans to a software environment requires that you explore and experiment with the possible solutions until the best one is found. So automating your support function is not a single-shot, ‘make-or-break’ project that can or should follow extremely rigid guidelines.
Instead, it’s a process of continuous improvement that requires collaboration between everyone involved. Crucially, to be able to automate your support at all, your team needs the ability to change requirements and try new ideas. Agile principles are critical to support a process that is by its very nature able to create and respond to change.
Start Mining (for data)
It’s difficult, not to say impossible, to properly automate a process which you don’t currently understand. If your implementation team doesn’t ‘get’ how customer support currently functions, and lacks historical data to back up human experience, then the chatbots won’t have a firm foundation.
From a ‘human first’ perspective, not involving the actual agents on the front lines can hamper the effectiveness of the product you want to implement (i.e. chatbots), and also the engagement needed from your people to ensure that bumps in the road are avoided.
Data may or may not be ‘the new oil,’ but it will certainly help identify trends and areas to focus on which were simply not obvious beforehand. Data is also crucial in helping fine-tune and train appropriate models to deal with specific issues. A prime example is support ticket data. The information we get from this data can help us understand what terminology is used to describe problems and can then be used to fine-tune the bots’ capabilities so they can correctly identify what users are talking about.
Core to the success of chatbots in support automation is having the right content to help users resolve their problems. Building a chatbot is not just about designing standalone conversations or simple dialogues. Preparing content needs to start as early as possible in the process, as the type of content needs to inform the design of your solution.
Here are some key points to consider:
- Audience: who is the chatbot being designed for?
- Capabilities: what should the chatbot be able to do?
- Interaction style: what style is most appropriate for different capabilities?
- Adaptability to context: should context in which conversations take place (location, user, past history, etc) influence the conversations?
- Platforms: on what platforms are the conversations taking place?
- Systems integration: Will we need integration with an identity service, a CRM, a support system, etc?
- Improvements: what systems should we put in place to support the improvement of the chatbot?
A chatbot can learn and improve over time by studying your users’ language and choices. Eventually you can enable ‘at-scale’ issue analysis and solution evolution; or in other words, you get a customer-facing workforce that knows how to solve problems at the same rate they come in from customers. This won’t happen overnight.
A chatbot is a natural solution for those organisations that need to automate areas of their customer support, but only when it appears ‘natural’ to users as well. To make sure you can rest easy and let the bots do the legwork, start by knowing your customers inside and out.