What is chatbot? Why chatbots are important? Can I build chatbot from scratch or use some chatbot platform? What are best strategies for leveraging chatbots for business growth?
Must be wondering — Is this for real? We know the idea of chatbots can be a little intimidating to begin with but robots aren’t here to take over our jobs or control humans, seriously!. They’re here to make our day to day life easier. The year of 2019 is set to become the year of the chatbot and businesses are gearing up themselves with chatbots to interact, engage and up-sell their customers at scale.
In this comprehensive guide, we’ll show you how to deploy smart chatbots for your business to free-up yourself from monotonous and time-consuming tasks. And raise customer experience to the whole new level by providing immediate assistance to your existing customers, potential clients and of course window shoppers to spread the word around.
Chatbot 🗣️ ˈCHatbät/ - aka virtual assistant or conversational agent - is a computer program based on predefined logic trying to emulate human speech or textual conversation. The entire experience is based on mimicking the real life conversation between two or more individuals. Today, such applications are widely seen in the customer service space, assuming roles conventionally performed by living, breathing human beings in the customer support or tele-calling department.
Chatbots leverage chat mediums like SMS, website and social messaging platforms like Facebook, Slack, Twitter, Whatsapp to receive and respond to messages.
From architecture perspective, a chatbot can look like a normal app. There’s an application layer, AI engine (optional), a database, and APIs to call external services except rich UI which is deliberately replaced by chat interface. On the contrary to mobile application which comes with rich UI interface having multiple tabs & drawers, chat interface is primarily governed by dialogue flow to allow peer to peer messaging focussing on what the user wants and how best to deliver that.
There are two kinds of modern chatbots.
There also exist third category — Hybrid chatbots having best of both worlds
You might have seen this conversation pattern in some of the chatbots. The primary goal is to guide the user in the process of getting the service he wants. The chatbot can simply start with few suggestions and let the user choose one. This is followed by series of questions in the attempt to narrow down the search further to acquire some user information while conversation continues. This one-on-one pattern continues until the chatbot has all the information to assist the user with the relevant information.
Rule based conversations might seem pretty basic to the user who’s expecting more but it’s actually designed not to bother the user as much as possible and stick to the business objectives all the time. Some businesses find rule based conversations very convenient to design with different scenarios good enough to engage the user and drive the conversation smartly.
The complexity is bit increased when chatbot allows user to enter input as free text message. At this point, bot expert has to make provision to ensure that chatbot only understands the input but also interprets it correctly. For instance when chatbot ask simple question to the user “what’s your email address?” In the ideal world one expects a person’s response with a valid email address but very often users unintentionally enter email address in the wrong format. To deal with such nuisance, some sort of validation is always required to ensure the correct input.
In short, from learning perspective, the rule based chatbots mostly follow linear conversational flow and knows how to handle answers, leaving less work for the chatbot owner to be done in terms of guess work.
Everybody knows Google Assistant or Amazon Alexa works. You may ask anything and expect some sort of reply which answer your query or in some cases it might prompt bot to ask few questions back to you in order to narrow down the scope of search.
Such conversational dialogues are trained with the help of Artificial intelligence. The bot making process for building AI chatbots is quite sophisticated. Broadly, it does always include these three protocols -
Data processing — Multi-step natural language processing workflow to correctly recognise and infer free input text. The process requires serious engineering efforts when input is multi-line &/or multi-objective. We’ve outlined some basic NLP steps required to process the raw text or voice (converted to text for NLP processing)
Contextuality — Availability of previous chat conversations (message logs) to build knowledge graph. This helps program to learn inter dependency between dialogues and the course of action so that chatbot can drive the conversation in the right direction.
Chatbot training — Provision to learn new conversations all the time to improve chatbot’s performance with the help of unsupervised methods based on machine learning techniques or in the supervised manner assisted by humans.
From learning perspective, it offers advantage to the user to initialise the conversation as per his or her need rather than guided by the bot in the conversation. Overall such user experience is quite reassuring that user has some sort of control over the conversation.
How to create your own bot? You might want to know the answer that how much does it cost to build a chatbot, well! it depends. There are three variables which affect the cost of chatbot development -
Business vertical — Companies developing chatbots for verticals like e-commerce, travel, fashion & hospitality usually look out for better customer engagement with robust design while government regulated sectors like health or finance (insurance) look out for data privacy protocols in place which adds up to significant overhead to the overall chatbot cost.
Type of interaction — If the chatbot is based on menu driven approach and uses pre-defined options for the user to select, the cost is usually low. But if the chatbot uses NLP methods to recognize the user utterance enabling humanlike interaction, it will leads to a higher cost.
Third party integration — For building simple lead generation flow or automating FAQs, the development cost is usually low. If you want your chatbot that books a flight or notify user on leaving abandoned cart, place an order, these chatbots require deeper integration with the backend system and will lead to a higher development cost.
Building chatbot from scratch is not a piece of cake. Having sound knowledge on how to mimic conversation programmatically, understanding utterances and training the chatbot to respond humanlike is a skilful work. It’s not like that you deploy the chatbot once and it will keep up and running on its own. It’s a full time job to maintain the chatbot with improved content and developed technology. Both time and money is required especially given it’s adoption.
Many a times using out of the box solution works better in terms of saving resources, especially when you want to immediately try out the product and understand customer or visitor’s behaviour. This gradually and eventually reduces customer support turn-around time.
Secondly, companies working chatbot technology keep adding new features so you will remain up to date in terms of technology. Platform providers also keep a tab on the method of conversation prevalent in the industry. They only use best trusted interactions and methods which are applicable for the wider population, covering the most common case scenarios.
If the current solutions aren’t sufficient enough to serve your needs, then you must talk to these companies for customisation. They will relatively do it in a better way catering to your specific needs.
Let’s grab a bag of chips! and be with us to learn about what brands and businesses need to look out so that chatbot conversation is fun, engaging and easy on hands both for newbie and veterans.
Before jumping on to design conversational flow, make sure you have figured out real problems that should be handled by chatbot. For instance — do you want your chatbot to be informational? Or do you want it to complete some task like collecting feedbacks or doing surveys? Ask yourself these questions and then start designing your chatbot.
Many a times user doesn’t know what to say. To take the take pressure off from the user, chatbot should provide some tips or suggestions to kick off the conversation as well as navigate the user throughout the conversation.
People like to respond to faces and that’s why chatbots should be personified to derive maximum attention. Choosing a avatar that could represent itself as face of the company, can play major role in shaping right perception for the brand. It is not necessary to come up with the avatar always, but something that can add some personality to the chatbot.
Pay close attention to the language and vocabulary of your chatbot that it uses to interact with users, as well as the info you can convey in association with the messages it sends — The choices you make here will shape your user’s perception of its persona, which sets the tone for the entire interaction.
Many companies are now considering chatbots as powerful medium to carry out meaningful interactions with their valuable customers — which proves the rich potential that chatbots offer in their simplicity and intuitiveness
We all have gone through miserable experience of filling bank, insurance, property and mutual fund forms. if you notice, you have to put all the mandatory inputs and even one column is missing or wrong the web page will prompt you to fill the entire form or at the least ask you to fill the required field again. This really frustrates the user at times and notable culprit for higher bounce rate. On the contrary, chatbot offers you whole new experience by mimicking human element in filling forms. Chatbot offers better way to funnel leads and interact with a customers online. Here are some top reasons why -
Suppose you are interacting with a chatbot regarding current offers on a specific make & model of the car — let’s say Hyundai Venue 2019. During the conversation it may ask you “Which colour do you like Red, White or Silver”? What type of engine do you prefer, petrol or diesel? Do you want to know the finance options on Hyundai cars? Are you planning to book a test drive?
You may not notice but the form is filling in the background with user preferences as and when user starts the chat flow. Steadily, the conversation derives you to the point when it asks for the personal information with human touch. So it’s like hitting two birds with one stone. There is a cherry on the top called “Partial leads”. What if someone leaves the conversation in the middle, still you have some information & self reported preferences which business can utilise for re-marketing campaigns on Google or Facebook. Bada Bing Bada Boom! 😃
Since chatbots are dynamic in nature allowing logical jumps or , it can smartly adjust the question based on previous response. For instance, if user has shown interest in Apple’s iPhone X. It tailors it’s next question based on previous response “Would you like to buy Airpods for your new iPhone X?”
Displaying questions to the user based on the initial response and subsequent responses makes entire experience surreal and human too. What we call this up-selling at it’s best. The experience that conventional web form may not be able to offer.
The last time you were on a live chat, did you get instant response to your question, or were you asked to drop your email address so that they can call back you later? More importantly, you may not have appreciated dropping an email address in the first go!
It’s annoying when you need an instant response to your query but you can’t get it. It’s frustrating!. You will either leave for another brand or send the email and expect a reply soon — which often takes more than 48 hours. It is sad, right?
Live chat is an awesome technology if implemented with ample human resources, it can help you move your visitors to the bottom of the sales funnel faster. However, when enough attention is not given to live chat due to lack of resources since it’s a costly affair after-all, it eventually leads to unhappy endings and delivers worst possible customer experience ever imagined by the business.
Another reason, GEN Y & GEN Z (millennials) act promptly and today’s most interfaces of big brands are still not able to cope up with these generations.
Although sophisticated chatbots can be very responsive, they are still limited to what they can do. In fact, it can’t give more answers beyond what it’s taught to say — come on, it’s programming.
So, is live chat winning the customer service game over chatbots?
We’ve gathered some top reasons why chatbot together with live chat can offer better engagement experience.
One of the top benefit of deploying chatbot is freeing up yourself from doing repetitive task like answering FAQs or helping customers with common case scenarios without involving human being. Psychologically, humans do find frustrating to answer the same set of queries again and again. This automatically saves resources and energy of business.
Chatbots are like Frontman for your business. They are the mouthpiece of your business, they converse with the customers and build trust for your brand. More importantly, they are not biased and user experience is always unified.
The 24x7 availability of chatbots makes them some sort of marvel’s comics character with super natural power. They never need a coffee break, never sleeps, never sick and never take a vacation. Wow!
A worth mentioning application would be implantation e-commerce chatbot serves thousands of customers simultaneously, making their shopping easy at any given point of time.
Most AI chatbots are programmed based on the decision trees to serve important business goals. Customer buying metrics such as time on landing page, visitation history, past preferences, items purchased and similar other insights form the basis upon which chatbot can take decisions autonomously in the best interest of the business.
Every business strives hard to save resources. Instead of recruiting, training and paying multiple agents for the customer support, chatbots offer excellent alternative way to deploy less resources and bring more automation to the overall process and improve productivity.
Are you sure your chatbot knows everything! You’re wrong. No matter how sophisticated your chatbot’s AI and machine learning capabilities are or how much attention you’ve paid to design it’s dialogue flow, there are moments when it makes a mistake. In some cases, these moments aren’t even the chatbot’s fault and are more so related to users not using the chatbot correctly or having false expectations regarding what it offers as a service.
During those critical moments, having a human agent as a backup will help business ensure that users are taken care of regardless who’s mistake it was.
Google trends allows you to see how popular search terms have been over time on Google along with related demographic information as estimated by Google. So, we used Google trends to find how popular “chatbot” term is —
You can see quite clearly that over the past several years especially after 2017 there’s been skyrocketing growth. You might notice slight dips but for the most part, the search volume for this niche is pretty stable.