After my introductory blog on why chatbots are essential to enterprises, here’s my second edition on how you could create chatbots yourself! As I had suggested earlier, here I am sharing with you the best practices you can follow while creating chatbots yourself.
Building chatbots have never been easier; all thanks to a large number of DIY platforms that offer step-wise descriptions detailing every point involved in their creation.
Ideation and planning are two critical phases that developers need to pay the most attention to while creating chatbots. Unless backed by thorough research, the chatbots may not prove to be useful to your organization and may remain as just another tool in the repertoire of organizational assets. Here are the factors, which you should consider as a part of the ideation and planning phase:
At the ideation stage, look for the answers to the essential ‘five Ws’ – who, what, where, when, and why. This will help in understanding the value proposition your bot will offer to your users. The biggest challenge that organizations face with their chatbots is chatbot adoption. Unless your users can see a value – a benefit – in adopting your chatbots, they will never be interested. For example, you need to be clear about what you expect the bot to do for your users; do you want your bot to only answer users’ queries or do you want to add extra call to actions (CTAs), such as to generate sales leads, make appointment bookings, show promotional offers, etc.
Right at the beginning, you also ought to lay in front of yourself a clear design roadmap stating where you want the chatbots’ conversations to start and/ or end; chatbots need a well-defined and documented design strategy to perform well.
Since chatbots appear as only-for-conversation tools, designers/ developers often ignore the design aspect. However, focusing on it can win you half the battle!
Using an existing framework will not only speed up the bot building experience but will also drastically reduce the bugs created during the development phase. Microsoft Bot framework and luis.ai are powerful frameworks to get you started.
Choosing the right technology stack is perhaps the most critical part of the ideation and planning phase. Bot development requires a reasonable amount of time, effort, and money – not considering the tech stack can negatively affect each resource. Microsoft Technology stack is the unanimous choice of bot developers; it is targeted to work with the Sitecore CMS as both are based on .Net technology.
In addition to the bot framework, Microsoft provides many tools for creating awesome bots, which can be further developed as per organizational needs.
You need to create a comprehensive workflow for your bot; it helps define the context of the bot and highlight its purpose. Focusing and finalizing the context(s) in which the bot will operate enables you to visualize the end-user’s actions while interacting with your bot.
For example, the flow for a bot that helps users to book a cab could be: finding a cab, booking it, confirmation and payment. However, the actual context is sharing location, confirming destination, checking a pending payment, checking the cab availability, and then finally offering a cab that the user can take.
Intent and entity extraction
Intent and entity extraction is for making the bot understand what the user wants. Natural Language Processing (NLP), Artificial Intelligence (AI), and Machine Learning (ML) are used for doing this. If it sounds overwhelming, there are a lot of vendors who have already done the hard work for you – including implementation of complex algorithms that go into machine learning, phrase extraction, user utterance – the only part left for you is to customize the pieces to your bots’ needs.
can understand the users’ thoughts and offer what they want, it can offer them (users) great convenience. It becomes handier if the bot can offer voice commands too in addition to textual intents; however, voice command-enabled bot development can become a bit tricky as the complexity of users’ thoughts generally increases in voice commands, and the bot will need to undergo a lot of training to understand and respond to voice commands.
Generally speaking, there are two classifications that bots can be grouped into.
In my experience in building purposeful bots I have found that building a bot without any flow, and only with the intent and entity extraction, can be the worst thing in bot development. Bot development without a clear flow often leads to users getting lost during the conversation, which eventually results in bad conversions and customer dissatisfaction.
Imagine, you were inquiring about the services that edynamic offers across domains. You get to select your organization from a list of options presented to you. The bot asks you to mention the size of your company but doesn’t provide you with a threshold, a lower and upper limit. This is because the developer of the bot didn’t anticipate this as a result of not thinking enough about the flow. So, all that the bot says is, “Sorry, I don’t recognize that”.
This will lead the conversation to a dead-end and compel the user to either start the conversation all over again or call it off.
These are bots, which do not handle intents and are completely focused on flow using CTAs, quick replies, and other quick-help links. These kinds of bots can turn out to be a significant source of disappointment to users. The flow needs to be defined in a crystal-clear manner right during the ideation phase itself.
Consider this: end-users typically enter simple phrases, such as “Which clients have you worked with lately”. If your bot doesn’t recognize this and replies with, ‘Sorry I have no answers for that’, it will undoubtedly create a wrong impression on the user and might result in the loss of a prospective customer.
Also, human psychology tells us that new services are adopted quickly only if it increases convenience and helps in getting things done via shortened paths. Generally speaking, ‘flow-only’ bots increase the path length for getting things done and have the potential of being a blocker in the end-user adoption of a newly-created bot.
Here are a few quick tips to create ‘Great’ bots:
At edynamic, we are developing powerful bots to meet organizational needs; we shall share the details with you soon. If you’re interested in a Beta release, write to us at firstname.lastname@example.org and we shall get in touch with you.