This chapter is dedicated to studying the business benefits of Azure chatbots by exploring different advantages of chatbots. We also look at measures used to demonstrate business benefits of chatbots. First, we examine some interesting statistics on chatbots to get a sense of their significance, before looking at the details of business benefits.
Percentage of the population in different countries using chatbots
As expected, the United States has made it to the top of the list. Following the United States, there is healthy competition coming from India, which is second on the list. India has the world’s second highest number of Internet users. Organizations are confident about using chatbots and they are fully harnessing the potential chatbots hold. Chatbots can be translated into any language. Germany is in third place, followed by the United Kingdom and Brazil. Not shown in Figure 6-1 are Spain, France, Canada, Italy, Australia, and the United Arab Emirates—all of which are similarly ranked.
Industries that use chatbots
One of the primary reasons chatbots are performing well in the real estate industry is because they generate plenty of leads. Chatbots today are widely adopted to produce leads, which is a necessity in the real estate industry. If there are no leads, there is nothing to sell.
The next industry on the list is the travel industry, which uses chatbots to answer user queries related to travel, thereby speeding up the travel booking process. A 24/7, 365-day presence of chatbots ensures that questions are addressed. The adoption of chatbots in education helps students engage in online courses, and helps students and parents complete the university admittance process. In the healthcare industry, chatbots are used to book appointments with doctors and diagnostic centers. Financial institutions are using chatbots to conduct preassessments and consultations with potential prospects.
Advantages of Chatbots
Chatbot benefits
Availability
Think about the occasions when you were required to wait several minutes before you could connect with a customer care representative to fix your issues, be they credit card related or insurance related, for example. There are then several minutes of waiting before your queries are actually resolved, which may be a painful experience. With chatbots, however, the waiting is over. Users can connect with chatbots whenever they want, in real time, to fix their issues. Chatbots are available 24/7, 365 days a year. They never get tired; they do not require breaks. They just continue to obey your commands and provide best-in-class user experiences. Some organizations have adopted smart chatbots, which means you can command them using speech and they will follow your orders.
Capacity Management
In the case of a call center, humans can communicate with one human at a time; the agent assists only one human caller. But with chatbots, this is a different story because they are able to communicate with multiple users. Regardless of the time of day or the number of people engaged in conversations with the chatbot, chatbots find answers immediately. There is no capacity constraint when it comes to chatbots.
Think about a restaurant that is famous for the quality of its food. In the evening, customer demand is high. It can be stressful to take food orders from customers manually, and sometimes restaurants are forced to deny certain orders because of capacity. However, when this restaurant starts using a chatbot, it is able to accept as many preorders and orders that it can, and serves its customers with greater satisfaction. If a customer orders a food item and that menu option is not available, the chatbot informs the customer immediately of the unavailability. Large food chains such as MacDonald’s, Wendy’s, Burger King, and Taco Bell have already installed chatbots to receive customer orders, and serve food with greater pride, independent of capacity.
Flexibility
When people move from one type of industry to another, they need to undergo training and acquire experience before they can start helping customers address their issues. This process takes time, and there is a monetary cost associated with it. The results may not guarantee success because employees may leave a company. In the case of chatbots, you can develop and design them so they can be used in any type of industry. This is possible by changing conversation flows with the use of relevant words connected to a database. Here the results are guaranteed, because chatbots cannot “quit.” When they are developed and designed, keeping the industry in mind, chatbots are all set to hit the floor and create a difference in your business. Chatbots are able to manage two-way conversations, which are quite prevalent in the customer service and retail industries. You can design chatbots that are aligned to industry type.
Greater User Experience
Humans are full of emotions and, sometimes, when addressing a customer with fairly large expectations, may affect the user experience negatively. Also, human-to-human interactions are bound to lead to conflict, most often requiring escalation to an upper echelon of people to tackle the situation. With chatbots, it is easy to manage customers of any type because chatbots are programmed to respond to customers based on the information fed to them and the rules built in them. Chatbots entertain customers with high, medium, and low expectations by meeting their requirements. Humans always prefer to have engaging conversations. When ordering from a restaurant for home delivery, a chatbot can give recommendations from previous orders, find your delivery address, and provide information on discounts and deals. The user is supersatisfied while placing the order, and the conversation is simple and engaging.
Low Cost
There are costs attached to hiring, training, and onboarding employees. In addition, the employee may not able to manage two or more customers at one time because the work is fairly complex, which may compromise the user experience. If an employee is unable to meet a sales target, this affects the business. Once developed, chatbots are able to engage thousands of customers at the same time, and users can jump into the conversation if they need to do so. Chatbots require a minimal development cost compared with hiring thousands of employees to support thousands of customers. One chatbot is able to serve thousands of customers with greater satisfaction and meet a monthly target. Even if targets are not met, there is not much of a business impact, but there are conversation trends, and organizations will have the opportunity to program chatbots in a new way to assist customers.
Faster Onboarding
If one is to accomplish a task with success, then one must train oneself or learn how to execute a task in the most efficient way to benefit an organization. Chatbots can help with the onboarding process Continuous training and teaching are involved at every level of the employee hierarchy that employees have to go through.
Employees come and go; this is a natural fact of business. Therefore, new employees must be trained. Companies want this to occur quickly; they want their people “up and running” as soon as possible so they are assets to the company. There is a greater possibly with chatbots that they could completely eliminate onboarding time, providing clear and easy-to-understand conversation flows with new employees. There is absolutely no doubt that there will be dynamic changes in chatbots too, but they will take little of your time to resolve compared to human employees.
Work Automation
Normally, people tend to get less productive when they are assigned to jobs that are repetitive. In general, we humans usually get bored by performing the same task over and over again. Chatbots have the capability of overcoming repetitive tasks by making them automated and executing them whenever they are required. There are already various Slack chatbots that are able to automate repetitive tasks, and this helps people save time and energy, and be more productive when coping with challenging tasks. Consider a situation in which you purchased an item and found it to be defective. You want to replace the item. There are e-commerce chatbots that do this with a single click. In addition, users can share images of the defective item with the vendor. There are smart financial chatbots that can inform you when there is flotation in stocks. There are smart chatbots that act as travel agents and are capable of booking your travel with a single click. There are two popular AI-based chatbots that have automated doctors’ work: Dr. A.I. by HealthTap and Melody by Baidu.
New Sales Channels
Today, chatbots are able to sell products online because they’re online 24/7, 365 days a year. In fact, during holidays, they are superbusy selling products because people connect to e-commerce sites more often during holidays. Per recent studies ( chatbotsmagazine.com) , in the digital world, 70% of people prefer texting (read: using a chatbot) rather than calling and placing an order. Because of the dominance of Amazon in e-commerce, there is no need to drive to the mall and make a purchase. You can find almost anything online and buy items with one click. Chatbots always have an opportunity to sell products based on the needs of each customer. Chatbots are capable of remembering a customer’s answers and can tailor their responses accordingly. By doing this, chatbots provide a personal touch and personal level of service that closely mirror human interaction.
E-commerce brands such as H&M, eBay shopbot, and Tommy Hilfiger are now selling their products using chatbots. Amazon has voice recognition AI (called Alexa) that is currently gaining in popularity.
Personal Assistance
Nowadays, people are able to use chatbots as personal fashion advisors to get recommendations for clothing, as financial advisors to request trading tips, as travel agents to suggest places to visit, as medical receptionists to book a doctor’s appointment, as a source of entertainment to book movie tickets, and on and on. As mentioned earlier in this book, by using personas, chatbots are able to store choices and offer relevant options the next time you visit. One well-know chatbot is the CNN chatbot, which helps users receive personalized news (Trim). In addition, a personal finance chatbot, Taylor, is a travel assistant.
At Maruti Techlabs, there exists a requirement-gathering chatbot, Specter, designed using the Microsoft Bot Framework. The objective of designing such a chatbot was to assist the sales and marketing teams in improving efficiency when sharing requirements. Normally, Specter asks users a few questions to understand their requirements. It then forwards the information to the technical team, which validates and qualifies the lead. This saves hours of going back and forth with different teams during meetings, e-mail exchanges, and telephone calls .
Disadvantages of Chatbots
Inability to understand: As a result of using static programs to design chatbots, sometimes they may get stuck if an unprogrammed question is asked by a user. This leads to a poor customer experience and creates loss. Sometimes, multiple messaging can also tax users and worsen the overall customer experience.
Complex interface: Chatbots are perceived as very complicated identities that requires plenty of time to comprehend user requirements.
Increased installation cost: As we seen so far, chatbots are pretty useful in helping you save a lot in terms of manual human effort by ensuring 24/7 availability and by being able to serve multiple clients at once. But, unlike humans, every chatbot is required to be programmed differently based on business needs, which often change, which then leads to increased installation costs. Furthermore, there is also an increase in the time needed to develop and deploy an updated program, and this should be planned with minimal impact on the business and on customers. There may be last-minute changes too. Therefore, additional development requirements will increase costs.
Time-consuming: Most of the time, the primary objective in deploying chatbots is to accelerate user response and create a better customer interaction and user experience. But, as a result of limitations such as data availability and the time required for a chatbot to self-update, any particular process is more time-consuming and expensive.
Zero decision making: It is evident that chatbots cannot make smart decisions as humans do. They are known for their infamous inability to make decisions. One large company experienced a situation in which its chatbot went on a racist rant. It is critical to manage the precise programing of your chatbot to prevent incidents that may affect your sales and reputation negatively.
Poor memory: Until we create personas in the background, which requires some additional investment, chatbots do not have the capability of memorizing past conversations, which forces users to retype the same conversation again and again. This can be extremely annoying for users. Therefore, it is important to be diligent while designing chatbots and ensure that programs are able to comprehend user queries and react accordingly.
To summarize, although chatbots are our future, we have yet to uncover their full potential. Their rising popularity means they will stay in the market for a long time. Machine learning has transformed the way companies communicate with their customers. Along with new technology platforms supporting the Microsoft Bot Framework, new types of chatbots are being developed. It is exciting to experience the development of a new domain in technology while surpassing previous ones.
Define Success Metrics for an Azure Chatbot
We have come to the last part of this chapter. Here, we define metrics that can be used to measure the success of Azure chatbots. A metric is a quantifiable measure used to track and assess the status of a specific business process and scope for improvement. Defining metrics is important because their use measures the success of a chatbot, and a return on investment can be calculated. For newly created chatbots, some of the metrics may not be applicable until the chatbot has been in use for a while. Companies need to monitor metrics closely. Most of the time, expectations for chatbots are greater conversion, efficiency, and faster response as they engage users. Companies must define the right metrics based on these key parameters. In this way, the performance of the chatbot can be monitored to improve its efficiency.
User Metrics
User metrics
Total users: Total users is the most basic metric. It records the number of people using your chatbot. This measure is important because trends and patterns will reflect a change in the number of users and the amount of data to which the chatbot has been exposed. This metric also provides critical information about market size, which may affect your chatbot.
Active users: The active users metric identifies the number of people who are able to read messages in the chatbot during conversations with users. Active users are potential targets for the success of a chatbot. The potential effects of a promotional campaign can be estimated from the number of active users. The number of people who read your message is absolutely critical. User engagement is not guaranteed, but the content is seen by users.
Engaged users: The engaged users metric identifies the number of users who enter into a conversation with a chatbot. They send and receive messages. The engaged user statistics is important because the chatbot will be able to provide the conversation statistics. This kind of metric definitely shapes decisions regarding the effectiveness of the chatbot. It will be pretty evident when a chatbot is unable to start a conversation with users.
New users: The new users metric captures the new set of users who saw messages in your chatbot after a promotional campaign. This metric is necessary to keep a count of active users. Customer preferences change over time, and the number of interactions with the chatbot may decline. For this reason, there is a need to have new users to keep your customer base strong.
Message Metrics
Conversation starter messages: Conversation starter messages are those kicked off by users, not the chatbot. This metric includes the total number of messages initiated by a user while starting a conversation with a chatbot. This figure is important for determining the organic growth of the chatbot. When a user initiates a conversation with a message, the chatbot can send a message to acknowledge the user message. However, companies work to reduce the number for this metric because it is more effective if the chatbot starts the conversation. When a business implements a chatbot for customer relations management or digital marketing, after an initial greeting is sent by the chatbot, it needs to continue to send messages to keep the user engaged.
Chatbot messages: The chabot messages metric includes the total number of messages sent by the chatbot during a conversation with the user. Other chatbot message information helps to measure the length of a conversation between a customer and the chatbot. In general, we want this number of messages to be high. But, the chatbot needs to adhere to one critical condition: send accurate messages during a user conversation. We have seen bad chatbot implementations when the bot cannot understand the user input. In these cases, the chatbot replies with similar words repeatedly.
Received messages: The received messages metric identifies the number of messages sent by a user during a conversation with the chatbot. With this metric, we can assess whether the user was engaged with the chatbot. If this value is low, we don’t need a chatbot. Ideally, in social media platforms such as Facebook and Twitter, this will make more sense and minimal sense using a chatbot living in Facebook or WhatsApp.
Missed messages: The missed messages metric is the total number of messages the chatbot could not process based on user input. This metric may be hard to calculate. The reason for not processing a request could be because the input string entered by the user was in a language the chatbot does not understand.
Total conversations: As its name suggests, the total conversations metric is the total number of conversations started and completed successfully on a given day. This leads to engaged users.
New conversations: Also as its name suggests, the new conversations metric is the total number of new conversations started. It includes conversations involving inexperienced users and those conversations initiated by returning users on a different matter.
Performance Metrics
Retention rate: The retention rate metric is defined as the percentage of users that returns to use the chatbot in a given time frame. This measurement is important because we need to keep customers engaged to extract insights related to their preferences by creating situations that make them spend maximal time engaging with a chatbot. Using these preferences, personas for each user are created and reused by the chatbot to respond to similar queries made by other users. High retention rates can be achieved by promotional campaigns, such as engaging with a chatbot to receive a 50% discount or guess a word to receive an exciting prize. These retention strategies can be achieved by a high-quality chatbot capable of meeting—and exceeding—customer expectations. Figure 6-6 shows a sample retention rate printout.
Sample retention rate printout
GCR of users scheduling a ride
Goal completion time or messages: As discussed several times in this book, chatbots need to provide efficient services and good user experiences; otherwise, there are plenty of chatbot substitutes, such as web pages or apps. Effort reduction to complete a goal can improve the user experience, which can be achieved in many ways. One way is to provide relevant answers to user queries on time.
Fallback rate (FBR) : No chatbot is perfect. It may fail because it cannot process user inputs, because of a design issue and so on. The FBR is the percentage of times the chatbot failed to process user input. A greater FBR indicates improvement is needed in the chatbot; a low FBR indicates better chatbot efficiency. This metric is very useful in the customer service industry, such as at help desks or call centers, where chatbots are expected to provide accurate information based on user input.
User satisfaction: The user satisfaction metric can be measured through exit surveys. At the end of a conversation, users can be prompted to rate their experience: Did this chatbot do well? Include a binary answer of yes or no. This metric helps companies understand the overall effectiveness of a chatbot as rated by the user.
Virality: The virality metric calculates whether an existing user was motivated by the chatbot to include other users in the conversation.
Summary
With this, we have now come to the end of this chapter. We looked at some important statistics regarding chatbots. We also examined the advantages and disadvantages of chatbots. Last, we looked at metrics for measuring the success of chatbots. In the next chapter, we look at creating new solutions using Azure chatbots.