October 17, 2017

Three Steps for a Designing a Successful Chatbot

Three Steps for a Designing Successful Chatbot

Chatbot technology is not new, but since 2016 the technology has moved into the forefront of innovation for many companies.  Many people have heard about Chatbots, but not many know what Chatbots entail or how they can benefit businesses.  The operation of a Chatbot is simple and easy as it echoes human to human conversation; the development of one is a complex process that requires many development levels.

The first such level is the product definition phase.  Defining the business goals, company needs and the final product is where every Chatbot developer needs to start.

Set out clear objectives

Chatbot Development Step 1

In the same way that every other project needs a set of defining goals, so does Chatbot development.  These goals should be definitive for the business (publisher) and the user (interlocutor).  Depending on the goals and objectives, the business and developer have to agree on the type of Chatbot to be built.  Is the Chatbot going to be task-oriented, conversation-led or functional?

Firstly, for any business with an online presence, the main objective is to drive traffic to the website.  For example, businesses can add an action button that directs website traffic to Messenger so they can engage with the Chatbot.  Real-time engagement means actualities in increased customers.

The next equally important objective is sales generation.  Sephora’s use of the reservation assistant Chatbot is an excellent working example.  Sephora reported an 11 percent increase in booking rates via the Chatbot app.

Customer service and support are essential for any company, more so for service-oriented businesses.  Yet, everyone can agree that it is time and resource consuming.  By using Chatbot technology companies are better able to manage these tasks.

As important as customer retention is, customer acquisition is more important.  As you cannot retain customers you do not have.  An important business goal is to engage new channels.

A major user goal is support.  User support can take many forms, from product assistance to company information.  A Chatbot interface should be accessible for all to address the various needs and customer or user support.

Seeking information, whether it is product information or a search for company contact details, access to this information should be easy and simple.  The Chatbot technology used should allow for easy and uncomplicated access to information.

Chatbots are problem-solving. Learning technologies that help you work smarter.  For example, the Chatbot available on Slack is able to match calendars and find free times for everyone.

Ultimately, your objectives should be grouped as follows:

Business goals:

  • Drive traffic to the website
  • Generate sales
  • Simplify customer support
  • Engage new channels

User goals:

  • Easy and fast support
  • Simple access to information
  • Quick problem solving

Define knowledge and domain

Chatbot Development Step 2

With the objectives set, step two is to determine the knowledge base of your bot.  What will it know, and how much will it know.

The developer needs to establish the knowledge breadth.  Clarity on the area of expertise is needed as it has a huge impact on our backend (UX).  For example, if you only have a set of predefined options, you have a simple developing task at hand.  On the flip side, if the search criteria are broad and cover a wide spectrum of industry knowledge, you will need a significant development effort that involves keywords, databases, and matching.

Establishing the depth of the knowledge base is a hard one.  The rule of thumb is to start with a basic feature set and then gradually improve from there.  As users are not completely familiar with bots and optimal conversational patterns are yet to be discovered, conversational backend still varies.

Familiarize yourself with bot speak

Chatbot Development Step 3

Once you decide to delve into the Chatbot world, there are a few new words and phrases you need to get used to

You do not need to be a super, hot developer to get started on Chatbots.  You can leverage off existing artificial intelligence (AI), datasets and Natural Language Processing (NLP) that work well.  Google, Facebook, and Amazon all have open and free NLP service to use for general purpose.  However, human conversation is nonlinear and the current AI available is not yet able to imitate actual human conversation.

One acronym you will encounter most often in NLP which stands for Natural Language ProcessingNLP is used to analyze the text input to determine what the user wants and then to deliver the relevant information or action.

Developers will also need to get to know the various platforms available for Chatbots as the chosen platform will determine your limitations.  You will also have very little control over the applications overall look and feel.

In order for a Chatbot to be useful, it will need to connect to data.  At the design phase, you will need to determine where the data is coming from, how the data will be delivered, the queries of each component and the relationship between the datasets.

This only covers the first part of Chatbot design.  As stated previously, developing a Chatbot is a lot more complicated and multileveled.  It starts with product definition and delves into further steps that include machine learning, hard programming, and AI options.

Leave a Reply