Chatbots as we know them are evolving. The generic bot found on many websites isn’t fit for many of the purposes it’s been deployed for. But now, chatbots are specialising to better meet the variety of roles they can fill – if correctly programmed.
By Ryan Falkenberg, co-CEO, digital self-service automation company Clevva
We’re seeing three different types of bots. The first is an information bot. These are the chatbots most people are familiar with. They can answer a query if the answer is documented and housed on a business’ website or in the knowledge base the bot has access to. Ask it anything nuanced or complex, however, and it gets stuck.
The information givers
In other words, these types of bots are fancy question answer machines that reference product and process manuals and FAQs to come up with defined answers to simple questions using natural language – which is where the sexiness factor came in.
The most brilliant information bot on the planet is the Google search engine. The difference between Google and your company’s chatbot is that billions of people are training Google and the content it searches is also provided by billions of people. It’s improving constantly because so many people are training it.
If you ask Google something, it spits out the top 100 answers and you tell it which is right and which is wrong, which trains it and it continues in a virtuous circle.
With customer service, if a chatbot gets it wrong, customers are less forgiving. They don’t want to train your information bot. They don’t want some generic answer or link to some brochure that requires them to read and work out the answer for themselves. They want their problem solved for them. That’s why most people hate information bots because they are seldom looking for information. Instead, they are looking for action.
The action takers
The second type of chatbot that has emerged is a transaction bot. Instead of just searching for the right information, this bot looks to action specific requests. You will have worked with one if you have tried to buy more data from your mobile service provider via their WhatsApp channel. As long as you are clear about the action you want performed e.g. ‘I want to buy 100MB of data’, these bots can help you.
Transaction bots are less capable when the required action is not clear e.g. if you say ‘I have run out of data. What must I do to get more?’. In cases where the bot first needs to clarify your request and analyse your situation, need or problem before a required action can be performed, they tend to require a live human to intervene and help.
This is because their brains are built to perform a series of specific and defined actions. They are not designed to first work out which action they should take. They are doers, not thinkers.
The query resolution specialists
The reality with customer service is that most people don’t have simple information or transactional requests they need help with. Usually they have a request, query, issue or complaint that needs a bit of clarifying before any action can get taken. What they need is a digital service expert, not a digital service assistant.
As a result, we are now seeing the emergence of a third type of chatbot – the service bot. Service bots are a significant step up from information and transact bots, which are built to react in a defined way to a specific request. Service bots are true product and process specialists, capable of helping each customer clarify their request, query, issue or complaint; analyse their situation, needs and/or problem; and then identify the right solution(s) before triggering the required actions.
To do this, service bots are built to handle queries in line with business and regulatory rules, while capable of adjusting dynamically to every customer’s specific context. Their brains don’t randomly self-learn. Instead, they offer you the assurance that every service query will get resolved in a consistent, complaint and hyper-personalised way.
Service bots are moving away from interacting with customers via the simple chatbot interfaces used by information and transact bots. Instead, they are using hybrid interfaces that combine the best of chat with the best of web app design.
This allows them to blend unstructured and structured conversations with richer web interfaces that surface information in multiple ways; gather data in multiple formats; and allow customers to easily move back in their journey to change direction, if required. As a result, service bots are able to deliver true one-touch customer service via your website, mobile app, WhatsApp and e-mail channels, without the need for a live agent to get involved.
Expect to hear a collective sigh of relief from your customers as they no longer have to endure the frustration of dealing with a limited information or transaction bot. Instead, they can have their requests, queries, issues and complaints resolved first time via the digital channel of their choice.
For most customers and companies, it could not happen quickly enough.