Recently, chatbots have become increasingly popular in business. Does this mean that every organization must have a chatbot that will interact with customers? In this situation, the main dilemma for a manager is whether the machine can be trusted to conduct a dialogue with clients or not.
At the current level of AI development, there are multiple types of chatbots. The most powerful of them can significantly affect the quality of customer service and a company’s reputation. Such bots – virtual agents – are able to recognize almost any user question, compare it with the information in the database, and give a correct answer.
In some sense, chatbots have become the next step in the digitalization of customer service, just like the emergence of companies’ websites and mobile apps were. Powerful bots are able to increase customer satisfaction because they are as good at consulting as people; however, unlike human operators, they are available 24/7.
However, some other factors need to be considered when deciding whether to implement a virtual assistant. Let’s take a closer look at the most basic ones.
Does a chatbot suit your company?
First of all, chatbots will be useful for companies focused on serving a large number of private (retail) clients. These are companies from finance, retail, telecom, transport, and tourism. When deciding to launch a virtual agent, first of all, it is necessary to estimate the volume of client traffic.
Companies working with a large flow of customers will benefit from chatbots more than others for the following reasons:
1. AI can learn from previous chat recordings, but for this purpose, hundreds of thousands of them are required. Training on a smaller number of recordings is ineffective, which means that the bot will remain “dumb”.
2. Launching, training, and testing a powerful chatbot can take months and cost a company more than a million dollars. Along with that, this will change the way the support/sales department works.
For some companies, such measures are justified. For example, Dish Network TV previously had to employ thousands of call center employees to answer the most basic customer questions. The introduction of the DiVA virtual assistant made it possible to immediately automate the answers to 30% of the most popular questions. Over time, this figure has grown to 40%, and the company continues to work on improving the agent.
Retail companies tried to introduce chatbots too, but with less success. For example, Amazon’s clients buy something using Alexa only in 2% of cases; as practice shows, a convenient application is often enough for purchasing. Equally, if a client has a difficult question, the virtual agent is often powerless, and the problem is solved only with the participation of a support employee.
Choosing the right channels
Voice platforms are enjoying the greatest popularity now. So far, about 100,000 additional applications for Alexa and several hundred thousand chatbots for Facebook Messenger have been created. These platforms are ideal for serving clients of some small companies. For example, Butterball launched its Alexa-based bot and managed to successfully process about 20,000 requests in just a few days.
Butterball’s service didn’t require client authentication, which is why it was so popular. However, if a company needs to work with a customer account and request their personal data, the attitude of users changes radically.
It is rather inconvenient for financial companies to organize automated support via services from Amazon, Facebook, etc. For example, Capital One has simplified the request procedure in their Alexa app as much as possible, but all customers can find out with the help of it is their balance and the list of recent transactions.
To ensure full support, banks need their own standalone application. This way, neither the company nor the client will have worries that confidential data will end up in the database of Amazon or another giant. For example, Bank of America’s chatbot runs on the company’s own application. As explained by the bank’s representatives, confidentiality is their priority, and they guarantee that no one will hear the conversation between the client and the bot.
The use of popular platforms is justified for companies with large inquiries, the answer to which doesn’t require the identification of the client by personal data. If customer requests are related to finance, health, or other confidential information, you can’t go without authentication, and in this case, the best option would be to develop your own application. If the company doesn’t have a powerful IT department, it is easier to outsource the development, implementation, and support of the application.
What to start with?
Before introducing the virtual agent, you need to prepare all systems that will be connected to the bot immediately after its launch. For a bot to work effectively from the very beginning, you need to provide it with customer data.
It’s recommended to start with small pilot projects. For instance, it can be a bot that advises on one particular service. As soon as the system proves to be effective, you can start attaching other products to it, up to and including the automation of support for all areas of the company’s activities.
As more data is collected, virtual assistants will become more efficient. Over time, such bots will have a good chance of displacing regular applications and platforms. Mass adoption of virtual agents is a matter of time, as they are able to serve customers faster than people who do it manually.