Everything you need know about ChatBots

The word chatbot is made up of “Chat” and “Bot”, it means chat powered and executed by bot (computer system). ChatBot actually emerged from “Chatterbots” back in 1990s. As it was too early to use them, the usage didn’t come to life until last decade. ChatBots were started to get used by organizations more effectively from 2015 onwards. In past 7 years the usage and industry for chatbots have grown by leaps and bounds.

A ChatBot is a computer software which helps organizations handle their customer queries automatically ay anytime (7×24 hours) anywhere (across the globe) using any device (mobile, tablet, laptop, PC, smart tv etc.) connected to the internet. 

ChatBots interact with diverse customer groups using messaging platforms which are designed to work and act as if a human customer service assistant is answering the queries based on keywords identified in the question and their model answers.

ChatBots require continuous update and learning as they are meant to answer customer queries related to product, organization, compliant, enquiry for information etc. There is also a need of having active human customer service agents behind the ChatBot to step in for situations where it is difficult for the ChatBot to answer further questions.

ChatBots are of many types, they can be very simple keyword based, rule-based Bots as well as sophisticated self-learning artificially intelligent (AI) Bots with built in natural language processing (NLP) capabilities as well as machine learning and deep learning abilities.

Here is a brief overview of Chatbot process and architecture,

ChatBot Process Overview and Logical Architecture

1.     The user generally opens a website or app to search for their need or to get support and guidance.

2.     The ChatBot popups with welcome greetings and either it starts by giving standard options that the user can choose to decide which area he needs help on or it allows the user to ask the query.

3.     User places the query which the ChatBot receives and processes. The processing is done through NLP (natural language processing) service consisting of NLU (natural language understanding) that helps understand users intent of the query based on keywords and phrases asked. The query is processed by the dialog manager and the response is given by Dialog manager to NLU (Natural Language Generation) which answers users query either in written or voice response.

4.     The user can see the response and ask further questions or after a waiting time of 1 minute the ChatBot can do the Thanks greeting to ask further questions and close the chat.

5.     In the background there are multiple technical architecture blocks and systems that operate to make the ChatBot work including storage systems, Knowledge base, API message Interfaces, Business Process Interfaces, IoT interfaces and ChatBot front end as well as back end conversational query engine.

We all are very familiar of ChatBots and use them in our daily lives. Here are some everyday examples of ChatBots including their types,

NLP Based ChatBot Overview

1.     AI based ChatBots – These chatbots are advanced and have NLP (natural language processing) capabilities. Some of them also have built in self learning abilities attached with machine learning and deep learning neural network algorithms.

a.     Siri on Apple Devices – The simplest form of AI ChatBot is NLP based Siri on Apple devices meant to serve our voice commands and answer our queries. Other than answering our queries, it is able to do many of our tasks, remember important events and remind us at the right time.

b.     Amazon Echo and Alexa – Alexa and Echo are voice enabled AI ChatBots and are enabled using NLP (Natural Language Processing) and AI capabilities. Other than answering our queries, it is able to search and order food, groceries and other items we need. Able to recommend the best price and online site to buy from.

c.     Google Home – Google Home is also voice enabled AI ChatBots and are enabled using NLP (Natural Language Processing) and AI capabilities. Other than answering our queries, it is able to search and book flights, movie tickets, hotels as well as suggest what is the best places to visit, with best time and price.

2.     Online Virtual Agent ChatBots – The other simpler form is the virtual assistant popup window we get on most of the websites greeting us and asking how can it help us. These are most common and seen across all business websites. Every sector and industry have their own ChatBots. E.g., Banking, HealthCare, eCommerce, Government Services, Logistics, Retail, social media etc.

3.     SMS Based ChatBots – There are also chatbots that are SMS based which are basically rule-based ChatBots that answer queries based on our questions and answers.

4.     Phone Call Based ChatBots – Phone call based chatbots are similar to SMS based chatbots but with voice enablement for communicating and handling user queries. During Covid Pandemic we have seen the use of these for managing online enquiries. We can also see them in use largely by banking, credit card and government services enquiry lines.

5.     Offline ChatBots – These are ChatBots used within organizations boundaries (E.g., HR, IT) and in some case for educational purposes by schools and universities. These ChatBots operate within specific knowledge domain to help manage queries from employees and students.

Now that we understand the type of ChatBots, Let’s understand the Pros (Benefits) and Cons (Downsides) of using them. 

Here are the benefits of using ChatBots,

1.     Improved 7×24 Customer Service – saving time and efforts for customers to find answers to their queries anytime anywhere with any device connected to internet.

2.     Standardized and automated processes – ChatBots make the respective customer service or query handling process fully optimized and automated for all possible scenarios. The process also undergoes continuous improvements and ChatBots continue to learn and feedback for corrections. Eventually making the organization more efficient and effective.

3.     Cost Savings – ChatBots can handle 1000s of customer queries at once using the same standard mechanism. This results in cost savings for the organizations’ operational costs as they will not need so many customer service agents working in multiple shifts and locations.

4.     Ability to Gather Customer Needs and Information – ChatBots can help gather customer needs and information through the questions and answers. The information gathered can then be used by sales, marketing and customer service divisions to engage with respective customers.

5.     Ability to generate leads – ChatBots can also help in generating leads on campaign sites and company websites by collecting interests and information of customers.

6.     Ability to Upsell and even cross sell – ChatBots can be setup to identify customer needs for products and services. Based on the identification, they can upsell and even cross sell products and services that can complement and better fit for the customer.

7.     Ability to sell extended warranty, accessories and CRP (customer replaceable parts) – ChatBots can be setup to identify customer needs for products and accessories. Based on the identification, they can sell extended warranty, accessories and parts to increase sell.

8.     Improved Compliance – ChatBots unlike humans are always well behaved in their answers to customer queries and every step of the discussion is recorded and audit log can be easily emailed to customers and stored in company logs for compliance purposes.

Here are some downsides and limitations of ChatBots,

1.     Limited Knowledge base – ChatBots can only handle limited set of queries and questions, as they are setup with a fixed set of keywords and questions identification and answers.

2.     Handle One Query at a time – In most cases ChatBots can only handle one question at a time. So if a user asks multiple questions the ChatBot will be able to answer only 1 question impacting user experience.

3.     Limited number of instances – ChatBots efficiency depends on the load and number of instances running parallelly. So organization must be vigilant to ramp up computing capacity for ChatBots based on their usage threshold.

4.     Complex Scenarios – There are complex scenarios in customer chat which ChatBot won’t be able to understand, answer and resolve. For such case the human service agents must keep a check on ChatBot discussions and step in to takeover such discussions.

5.     Need for Human Agent – It is normal that most humans would like to speak to a human services agent instead of ChatBots as conversing with ChatBots could take longer time especially on SMS and Phone based conversations.

6.     Language and Slangs – ChatBots understand normal language with correct grammar and slangs. They would not recognize certain slangs and language shortcuts that we normally use as humans.

7.     Over Loading – ChatBots can be attacked and kept loaded by hacking bots putting a lot of load on system capacity by opening many instances and asking many types of questions. Eventually impacting efficiency and effectiveness of the system and in some cases even crashing the system.

8.     Hacking – ChatBot platforms can be prone to hacking and hackers can alter the answers and language which can impact organization’s brand image as ChatBots are front face of organizations online presence.

There are many ChatBot software and platforms. Some of the famous ones are Hubspot, Zendesk, Intercom, Tidio, Mobile Monkey, Drift, IBM Watson and QlikSense. In addition, social media platforms like Facebook (Meta) also offer their rule based chatbots for business pages and ecommerce. Similarly, eCommerce Hubs like Shopee also offer their own ChatBot for online shops running on their own platforms. The top 3 Cloud providers Amazon (Amazon ChatBot), Microsoft Azure (Azure ChatBot) and Google (Google Dialog Flow) also have their own pay for use ChatBot platform offerings that can be easily built-up using scenario-based conversation flow configurations without any need of complex coding and development.

ChatBots are very much in use across the world by all industries and sectors. They are here to stay and they will become much more artificially intelligent in the upcoming years. AI based ChatBots combined with Deep Learning will open many more opportunities for their use to make our lives far more advanced, efficient and effective.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s