According to a recent Digital 2020 report, 56% of internet connections are made from mobile devices, and 89% of their time online is spent on instant messengers and social media. This is where businesses should look for customers, target their marketing and get feedback, and the most efficient way of doing so is through using a chatbot. A chatbot is an artificial intelligence software application for automated communication with the user, based on machine learning and built into a messenger or a social media platform.
Simpler robots use ready-made scripts: they respond with predefined text retrieved from a knowledge base. On the other hand, AI bots can teach themselves. Neural networks can be taught to draw their own conclusions, and the more an AI bot talks with humans, the smarter it becomes.
Chatbots can serve customers by helping them select and order goods, search for information, and provide consultation. They can play the role of a planner and even a friend and adviser (AI enables even some emotion recognition). Chatbots also greatly improve customer service efficiency because they do not want a salary, do not need a sick leave or rest, never have a bad mood, and learn on their own.
Here are a few examples of how bots benefit companies in different areas of business.
IBM sells software, computer hardware and IT services. Like any other large company, it has a support service. So the corporation introduced its IT support help-me conversation Watson chatbot, which replaced live operators taking over 80% of dialogues with customers. Thanks to artificial intelligence, Watson assistant can convert speech to text and text to speech; it can ask disambiguating questions and provide the answers it has found in the database.
Interestingly, due to self-learning, the system can process natural language to extract the meaning. The customer does not need to use exactly predetermined phrases to be understood, as it happens with Siri, for example. By analyzing the content and format of the questions, the bot can even identify who it is talking to – a professional user or a “dummy.”
Watson uses this data to choose the tone of communication – whether its response can include specialized terms for advanced users or an explanation that is as simple as it can be.
Customers turn to robots if they have a problem with a product or when they don’t know what to choose. If a job is too hard for Watson, it will re-direct you to a human operator. You can even talk to the virtual assistant about unrelated issues or share good news.
Autodesk is software for 3D-modeling of buildings and infrastructure, drawing designs and models. At first, customers requested help when they could not figure out the program’s interface. With the introduction of paid subscription, the workload for Autodesk customer support increased multiple times. If a user made at least one mistake when entering his or her data in a registration or authentication form, the software would stop working. Customer support managers could not handle the avalanche of calls.
Autodesk entrusted Watson with solving customers’ problems. The artificial intelligence learned from the information in chats, forums and manuals. It analyzed terms and keywords, idioms and colloquial expressions to understand nuances of the natural language. The chatbot was not only taught to recognize the meaning of words but also subtle connotations and implications. For example, “how much credit do I have?” is not just a question but an unspoken intention to buy. Watson learned to not just reply but to anticipate users’ intentions. Before the smart assistant, Autodesk took a day and a half to fix a problem; now it only needed 5 to 10 minutes.
Watson can help not only IT experts, but also medical personnel. Russia has a similar system, Doc+. The robot obtains the medical history of patients and asks clarifying questions. The information is then recorded in the patient’s medical history sheet and sent to the doctor prior to the consultation.
In Germany, patients use the Florence health assistant. The user uploads their assigned therapy and Florence reminds them to take pills or do a procedure; it also monitors the daily regimen, provides information about diseases, drugs, local pharmacies and makes a doctor’s appointments.
The Pyaterochka supermarket chain uses a bot that helps to find and recruit employees. The program was developed by the Stafory company from St Petersburg. An HR manager tells the robot what skills the employee should have. The system analyzes job search sites, picks suitable candidates, calls them and tells them about jobs at Pyaterochka. If the person likes the offer, the chatbot connects them with the HR department.
Using chatbots in messengers, candidates learn about the company’s vacancies and get interview appointments. The virtual assistant draws them a map with the stores located nearby with regard to the address in the CV or the user’s location data.
According to Pyaterochka HR director, it took from one to three weeks to find an employee; with the chatbot, it takes just 5-6 days.
Belarus-based company SmartSub developed a smart marketing bot that creates a landing page and contextual ads, sends analytical data, and increases performance indicators. The bot uses a chat interface to interacts with users; the latter do not have to visit any site or see into Google AdWords or launching ads.
The software was designed specifically for small business owners who just started using the internet and do not have a clue about online promotion while lacking enough finances to hire a marketing expert. The chatbot was initially created to be used with the Yandex.Direct contextual ads system but the software has been rewritten for Google and is now being sold in the US market.
Selling goods and services
During the process of interaction utilized by the Line Works interior design studio between a client and the bot, developed by DDL Bot, a user watches a video on Instagram and clicks the See More button. The chatbot offers taking a look at the studio’s works and selecting an interior design project. 90% of users agree and continue to communicate with the chatbot. When a person responds, the virtual assistant receives his phone number as they communicate via a messenger.
Then the system receives information about a client’s aims and preferences by asking him/her a number of questions. The robot learns about the apartment type, layout, area, materials and preferable style, as well as the client’s budget, and asks to provide an email to send an interior design project. All information about the potential client is forwarded to the studio’s customer relationship management (CRM).
Now the company’s manager can instantly contact the client and close the deal. In case a client refuses to purchase the service right away, the virtual assistant suggests sending a series of letters that contain the company’s business cases. Unlike emails, such letters always reach potential clients, with 98% of them actually opening the mail.
By Christina Firsova