An increasing number of businesses have added chatbots to their digital strategies.
While more bot APIs and development tools are emerging, questions around best practice for conversational UI design, maintaining brand integrity and the benefits of the platform for business are on the rise. Here, we answer the most asked questions from our clients around chatbots, conversational experiences and what it means for business.
“What is a chatbot and how can it benefit my business?”
A chatbot is a piece of software built to simulate a one to one conversation with your customers and offer them a specific service. Integrated into existing messaging applications, bot software use artificial intelligence, clever conversational design and natural learning processors to guide users to a solution through the medium of ‘chat’.
When done correctly, chatbots can act as a huge differentiator for business. Chatbots provide an always available presence across multiple text channels like WhatsApp, Slack and Facebook Messenger, speaking directly to customers or employees where they are in a hyper-personalised way. This opportunity to have one to one contextual conversations with customers meets the ever growing want for individual communication from audiences like millennials whose values are driven by immediacy.
“How do I do it well?”
While it seems like Chatbots are a relatively new way of communicating with customers, it might surprise you to know that Facebook Messenger alone has over 34,000 chatbots available to its users as of late last year. While this shows how many businesses have already developed a conversational platform, it also proves how difficult it is to get right - much of these 34,000 chatbots are not used.
The key enablers to success is to have engaging, actionable and contextual conversations which are functional but flexible, always responding to interactions, whether they are valid or not. The conversations should build upon the context gained with each interaction, with every question having a purpose and value to your user so they know they are nearing the solution they came for.
As with all development projects, it is important to get the scope of your chatbot defined, asking questions like why is your user interacting with a messaging service and how can you solve their problems in this channel. Starting with a design sprint is a good way to understand the needs of the user and how a chatbot can meet those needs effectively.
“How do you design a conversation?”
A conversational UI is one that mimics human interactions rather than following traditional computer interfaces. A conversation with a bot in its most basic form can be thought of like a tree - beginning with a welcome message and branching to different contexts, depending on the user’s interaction. More advanced conversational UIs use AI and machine learning to dynamically construct the user journey based on the user’s responses.
Like any design project, there is no one size fits all design solution. There are, however, some conventions which can be used as a standard.
As with all design, but more critically in designing a conversation, the interface should acknowledge human emotion.
Friendly and inclusive language should be used alongside acknowledgements and clarification with each response.
The messages should be digestible - no blocks of text should be sent by a chatbot.
The conversation’s flow should reflect as many potential use cases as possible, including an exit strategy and a method for handling invalid information.
The conversation should progress with each interaction, with value at each point to keep the user engaged and active in the interaction. In fact, if the platform allows, using buttons with quick responses is a great way to keep a conversation pointed and allow for specific flows through to a solution.
“How do I maintain my brand in a chatbot?”
A valid concern for many companies who are thinking of creating a chatbot is maintaining brand integrity. Chatbots live in existing messaging applications like Facebook Messenger, WhatsApp and WeChat, which may seem restricting when it comes to business differentiation.
While your businesses may not have control over the platform’s UI, it does have control over the conversational UI, bot graphics and name, tone and language. These are the key components to brand integrity when developing a chatbot.
The language and tone of a business’s chatbot should reflect the company's existing guidelines, while also being reflective of the problem which the bot aims to solve. Using existing platforms can also be seen as a positive, as it allows a business to understand exactly how the customer is interacting with the application and their level of expectation from their interactions. It aids personalisation, as platforms like Facebook Messenger, Slack and WeChat already have a pool of information from your user, giving the conversation instant context. This can be used to remove any friction throughout the conversation, allowing you create the best customer experience for your user.
“What platform should I choose?”
It is important to meet your users where they currently are. For example, a bot for enterprise may be better placed on Slack than on Facebook Messenger. The platform is also dependent on the scope of the chatbot and the service it aims to provide. Discovering where your user is and defining your chatbots services are both key parts in deciding on the best platform for your bot.
While some platforms allow for quick responses and ways around conversational breakdown, natural language processors (NLP) allow for conversations that mimic a more genuine conversation. NLP are built into a lot of bot platforms which power development, which makes it far more accessible to businesses than before. However, companies should choose a platform that will allow for the right balance between railed conversations (preloaded responses or ‘answer yes or no’) and more organic conversations that can be interpreted by a processor. If your bot is providing a service that requires a more organic interaction, like a bot that is providing legal advice, then the balance will lean more towards NLP. If it is a more basic chatbot or a pointed conversation, like a bot for ordering pizza, being clever with the flow of the conversation and designing it to use predictable responses is the best approach.