How viable are chat interfaces?
BY JACK JOHNSON
This is the 2nd article in a 2-part series exploring chatbots as an experience. If you haven’t read the first article where we look at how chatbots can be used for high volume interactions, you can find the article here. But how viable are chat interfaces?
The chatbot market
The chatbot market had an estimated worth of $190 million during 2016 and is predicted to see significant growth to $1.25 billion by 2025 which shows an unprecedented rise in adoption on a large scale. According to market analysis submitted by Grand Research Report, the top 3 markets expected to adopt this tech are E-commerce, Healthcare and Banking financial services. Although this market growth looks promising, it does not necessarily quantify whether the interfaces are viable for everybody or even at all.
Is it for you?
Due to the sheer number of use cases among chatbots and companies, it’s hard to attach a specific objective to chatbots. This makes it difficult for companies to determine whether chatbots are a viable solution for them, although researching use cases can help provide an outline. Implementing a chatbot needs to make sense for your company and your customers. Brooke Robinson a professional marketer claims: “as a brand who is ready to invest in the development of a chatbot, you need to consider if the build will enhance the consumer experience and/or drive business efficiencies.”
Businesses may look at this tech as a solution to a problem they don’t have. Meaning they implement the tech and risk providing a bad customer experience as the need is inadmissible. This, in turn, may disconnect them from their customers. Prior to jumping into chat interfaces, companies should have a clear outline of the flow they wish to automate. For example, an embedded bot within your website enabling users to swiftly find information that may be challenging to find (shown below) in large website structures. Or allow customers to purchase goods or services through a social channel like UBER. Identifying these will help companies better tailor the experience to their customers in a meaningful way.
A chatbot like any other application needs to integrate into a chosen channel, this can be through backend or social channels. Most traditional integration involves social channels including Slack, Kik and Facebook Messenger, it is these bots that will typically communicate within a rule set created by the company. The majority of social channels such as Facebook provide frameworks and SDKs (Software Development Kits) to help developers build integrations. Through the use of available and open-source APIs (Application Programming Interface), it’s possible to communicate with bots from within websites and apps.
For a predetermined workflow to become automated by a bot it is crucial they are supplied with an educated foundation to be able to resolve customer requests in an effective way. This foundation is based on information fed into the bot which after deployment will need optimization to help finetune the experience. For a rule-based chatbot, it will answer questions based on an outlined rule set. These bots can deal with simple queries but struggle with anything complex, it is essentially only as smart as it is programmed to be. Self-learning bots can either analyse and retrieve the best answer to a query or generate an answer making them more efficient. The workflows for those bots described work somewhat similarly.
This diagram shows what a typical chatbot workflow looks like:
Traditionally to trigger an interaction the user will send a message to the interface, although in some cases the interface will message first asking the users intent. Once this has been acquired it will be forwarded on to the NLP (natural language processing engine)(2). This engine will extract the user’s entity, intent and context (3). This information will then trigger a search of the database for a suitable response to the initial query (4), this is then handed back to the chatbot engine (5). The information is then collated and sent back as a structured answer to the user (6).
As with any other technology, chatbot security is of the utmost importance. Even more so as chatbots are equipped to handle financial and personal details through social media platforms bypassing rigorous verification processes. Although this worry is understandable chatbots are equipped with varying layers of security to help ensure their data is secure. Chatbots, for the most part, don’t really present security issues that haven’t already been fleshed out. While they are a fairly new technology the protocols used to protect them are not. The hosting platforms have their own security systems in place.
Chatbots use two main processes to ensure data protection. The first is user authentication and the second is authorization. Authentication can include two-factor and biometric. Much like banking apps when the user’s identity has been verified a generated authentication token will be sent to the receiving chatbot along with the request, in turn, they will then provide the requested information back to the user. For added protection authentication timeouts can be implemented which will revoke the user’s access after a given amount of time.
Providing authorization usually requires the user to undertake a specific task. For example, a push notification sent to their phone instructing them to follow a link or re-enter a code into the chat interface.
Chatbots can add substantial value to customer interactions at a low cost offering exceptional results. It is important to note that although chatbots are extremely effective they are not a replacement for a human service agent and should be kept in mind. The points outlined within this article should be considered before opting to adopt this technology. To conclude, chatbots are a very viable solution for companies looking to resolve high volume interactions.
We help teams across the UK deliver successful digital projects. If you’d like to have a conversation with us, we’ll put the kettle on.