What Makes a Great Chatbot? (Part II)
Non-functional chatbot features
Artificial Solutions and Botium have partnered up to create a brand-new guide for those looking to design and develop highly intelligent industry-leading chatbots. We’ve already learned about the most important functional chatbot features in Part 1, which you can read here.
Now, we’re ready to take a deep dive into non-functional features, which are just as important. Non-functional features are specifications that describe the system’s operation capabilities and constraints that enhance its functionality. These features contribute indirectly to the user experience, but they can hardly exert their power without the functional features.
Your chatbot could comply with the highest security standards if the chatbot cannot handle a single task in hand. But to not diminish their importance, once the basics are in place, non-functional features can be game changers. (PS: all the features discussed below and in Part 1 are available on Teneo, our industry-leading chatbot building SaaS (Software as a Service) platform!
Chatbot Platform Security
Since chatbots offer a wide range of applications, in certain cases they become responsible for collecting and protecting personal information as well.
The responsibility of ensuring chatbot security has become more pronounced after the introduction of GDPR in Europe. According to the regulation, the provider must be aware of a data breach and inform all concerned people in the upcoming 72 hours (about 3 days). Such leakage can lead (and have led) to a serious impact on the affected individual and the company itself.
The Security Testing included in Botium Box uncovers the vulnerabilities of the chatbot and determines that the shared personal data is protected from possible intruders. As for Teneo, all the conversational data is owned by your company, remains private and is not made available to other parties. Also, the ability to obfuscate PII data for GDPR compliance using named entity recognizers is available.
Today, personalization is the key to a good experience. Although it is impossible to develop a separate chatbot for everyone, it is feasible to build one that learns from the user history and drives the conversation accordingly. Chatbots have the potential to store the information offered by users and turn it into personalised recommendations.
Imagine asking for a refund from a customer support chatbot that cannot serve you with an immediate answer, but can actually help you the next time you enter the question: “What about my refund?”. This of course requires the chatbot to remember the previous conversation and to fetch the necessary information from the CRM.
This is also a great way to know your customers better and to apply cross-selling techniques once you learn their preferences. For that, multiple resources like a carousel, buttons, and flows can be used.
Chatbot Learning ability
There are more and more business applications where chatbots with self-learning capabilities can interact with humans. The learning ability of a chatbot refers to the competence of a system to learn from previous inputs. One of the ways they achieve this is through natural language processing (NLP), which enables the chatbot to continually get better at answering those questions in the future.
Behind the big power of the learning ability of a chatbot, there’s also the risk, given the potential for bots to be “reprogrammed” by users. Some businesses deploy a half-baked chatbot and end up training it on their customers.
NLP Testing in Botium Box helps to leverage the learning ability of your chatbot over time and to adapt this knowledge into context. Teneo comes out-of-the-box with fully configured machine learning (ML) capabilities. These cover the core optimized ML algorithms, a UI to create classes and maintain training data and in-built model testing. Once the system is running, Teneo constantly monitors ML performance looking for possible conflicts, intelligently suggesting improvements to the developer and highlighting suitable training data to increase accuracy.
Accessibility of a Chatbot Platform
The accessibility of a chatbot should be an important aspect in order to benefit more customers, reach a wider audience, and prevent complaints or lawsuits.
The presence of a button to access the bot is enough for some users to understand that a chatbot is available on the page. For people with disabilities, it can still present a wide range of difficulties.
Voice bots can easily overcome the challenge of accessibility for visually impaired users, but text-based bots must cope with this problem uniquely. Screen readers can support such applications to be able to notify users of the context of chatbot conversations and help businesses to meet accessibility regulations.
What about you? Can you think of any other features that should be in this list?
Remember, if you didn’t read Part 1 of this guide, you can find it here!