Chatbots have come a long way in the last few years and advances in AI, Natural Language Processing (NLP) and Machine Learning (ML) have brought along a huge transformation in the options available to handle customer support. These advances bring with them several benefits: chatbots that leverage conversational AI can facilitate more personalised and empathetic service, faster response times, and they can present relevant, accurate information to customers in the moments when it is most needed; even when companies are dealing with high volumes of enquiries that have variable-dependent answers.
These chatbots are designed to relieve bottlenecks in customer support centres that cause customers to wait in long queues before their questions can be answered, and provide better access to call centre agents when customers have queries that are too complex to be answered by the chat. As more and more of our business moves online, it’s more important than ever to build smart, scalable solutions to support your customers. You may be ready to graduate from a live chat agent to automated assistance, or perhaps you’re skipping straight to AI to speed up the process. Here are a couple of considerations you should keep in mind as you decide which solution will best suit your business.
Chatbots take multiple forms. We recommend putting your customer at the centre of the discussion when determining which one will best suit your business’s needs. What type of service will best benefit the customer? How will they use this new service? Here are a few examples:
Rule-Based Chatbots: This option is the simplest to implement, and doesn’t require AI - but the result can only perform a specific task. These chatbots are useful if your customers need guidance performing a pre-set task, like choosing the best purchasing option from a selection of similar items. Apple uses a rule-based chatbot to automate the process of helping users select the correct size wristband when purchasing an Apple watch, for example.
Pro: Help guide customers through specific tasks; automated. These are often out-of-the-box solutions.
Con: Rules are pre-set - these chatbots won’t be able to assist with queries outside of the pre-set flow: customers are likely to encounter a “Sorry, I don’t understand” answer from a rule-based chatbot.
NLP Chatbots: Natural Language Processing can be used to add conversational intelligence to chatbots. This option does use AI, and the resulting chatbot can understand the sentiment and intent of a customer’s query. They can identify the best information to share with a customer with a question like “What type of car insurance coverage should I buy for my teenager?” or a student asking “How do I apply for scholarships next semester?” - even though these answers have multiple dependencies.
Pro: These chatbots understand your customer’s needs and can identify and present the best solution. They can identify chats that need to be escalated to a human agent, while efficiently answering other queries.
Con: The best NLP chatbots are built using your business data, so there will be some waiting time during implementation and training.
Virtual Assistants: If you’ve interacted with Siri or Alexa, you’re already familiar with this type of chatbot. These assistants are designed to assist a human with a task when prompted. In the case of Alexa and Siri, these tasks include searching the internet, adding a reminder, or setting a timer. For your business, they may be tasks like applying for a mortgage or completing learning modules for an online course. Clearhead, for example, offers a virtual assistant focused on helping users track their mental health, and can identify the need to connect users with a human therapist when needed. Our client, Instamortgage, employs a virtual assistant named Rachel to help customers through the process of applying for a mortgage.
Pro: Virtual assistants can help your customers with a variety of complex tasks, alleviating stress from your operational teams, improving customer experience and speed to result. They can also be connected to an ML engine to perform business operations like claims processing or gathering information for a quote.
Con: These assistants can be complex to set up, and often require integration from multiple systems. Engaging with an AI partner that can identify and implement the best solution for your needs will be critical to success.
You may already have a pretty good understanding of what type of chatbot you’d like to implement. Before you proceed, here are a few more things to consider before you begin the process of implementing your chatbot.
Conversational AI is the difference between a simple, pre-scripted conversation and the ability to recognise intent and empathise with your customers. Although some interactions are easy to script, you can transform the way customers think about and interact with your business by providing automated support that truly understands their needs. Conversational AI allows for a semantic understanding of customer queries, and it identifies the meaning in customer queries rather than triggering pre-populated responses based on keywords.
In many cases, Conversational AI chatbots benefit from the presence of a specialist conversation designer on the implementation team. Conversational designers consider how the customer will interact with the AI, and anticipate things like slang, passive aggressive statements, and customers who can’t quite explain what they’re looking for. This way, they can ensure that the discussion can be orchestrated in a natural, straightforward way.
Your business’s conversational AI experience can be built on a variety of platforms, including IBM Watson Assistant, Google Dialogflow, and Amazon Lex. Spark 64 is technology agnostic, which means we work with all of these platforms (and more) to ensure success. Each option has its own benefits, and we’ll work with you to identify the best solution for your unique business case.
You can learn more about Conversational AI platforms by reading our white paper.
How common and straightforward is the customer journey for your business? An extremely straightforward customer journey like shopping for shoes online most likely doesn’t require Conversational AI - a rules-based chatbot that can offer direct information about shoe sizes might suffice. However - if the journey is a little more complicated (are your customers applying for mortgages, insurance claims, or university?) then anticipating their needs can be critical. Your chatbot may open with a different question depending on what page the user is currently on, or pages they’ve visited leading up to their query.
The chatbot can also gather information about the user to help identify what information will be most relevant to present to the user - for example, by asking an user who is interested in purchasing car insurance for information about their vehicle and driving habits.
One of the most common types of chatbot implementations we see is designed to identify answers to customer queries from complex knowledge bases and serve up accurate information. This is a worthy implementation, but it only scratches the surface of conversational AI capability. Your chatbot can be integrated with existing CRM and ticketing tools along with other internal systems so that they can actively assist with queries that require account information (such as “What’s my account balance?” and “when is my next payment due?”) as well as track the status of support tickets - all without the help of a human agent.
In the last few years as COVID-19 has dramatically changed the landscape of customer interaction, providing the most up-to-date information to customers can be a challenge. One of our clients, a large university, uses a chatbot to assist with student queries. They quickly realised that communicating accurate, updated information about campus closures and protocols was critical as COVID situations changed so rapidly.
Spark 64 helped to build a system that allowed their contact centre team to update the knowledge base without code so that they could get changing information out to students quickly.
Designing a solution that allows for ease of access to update information can be a critical consideration in the long-term success of a conversational AI project.
Any AI automation project should be viewed during the lens of customer experience. By prioritising projects that will positively impact this experience, you’ll be able to drive meaningful, quantifiable improvements to your business and create lasting customer relationships. There possibilities of chatbot implementation are nearly limitless - from simple bots that answer frequently asked questions all the way up to complex systems that can assist with mortgage applications, insurance claims, and even digital humans that can empathise with your customers.
Whichever option you choose, the best automations you can introduce are the ones that work seamlessly with the rest of your business. If you’d like to learn more about our work, and be the first to hear about innovations in the AI space, sign up for our newsletter below!
Chatbot terminology can be overwhelming for individuals without a technical background. This guide breaks down the concepts associated with this emerging customer service tool.
The story of how Team Humble Wisdom's Hackathon project - the Imposter Bot - came to life. Written by team captain and fearless leader Ming Cheuk, CTO of Spark 64