Bots, as one of the most promising artificial intelligence technologies, allow to enhance business efficiency and improve customer communications. This issue was revealed at AI Conference Kyiv by Aleksey Skripka, the founder and owner of EVE.calls.

The key ways to increase the performance of employees allow to divide tasks into subtasks and then delegate each subtask to certain team members. By the way, Henry Ford used such an approach in the early twentieth century and conducted a car industry revolution. Aleksey Skripka believes that modern technologies can lead to the next workflow revolution.

Modern trend: voice bots for businesses

According to the speaker, his company was exploring the automation and artificial intelligence market with an eye for trends three years ago.

“A chatbot trend initially became popular in the USA and consequently reached the CIS countries and Ukraine. We were also going to engage in chatbots, but we analyzed the audience and came to the conclusion of its preferable communication channels,” Aleksey Skripka said.

The analysis shows that 50-60% of users prefer a voice channel. It means that the most productive contact with customers is established by the phone.

This fact has a range of reasons:

• the telephone service is a familiar communication channel;
• it is fast: you can resolve any issue in a minute unlike email correspondence and chat communication where you should wait for the answer;
• one can provide more information using a voice than using a correspondence within the same period.

To help companies receive greater profit and save significant sums, they have developed a voice bot called Eve. It relieves call center staff and performs a half of their routine work, saving company’s time and money.

A typical call structure:

• call completion or automatic redialing;
• greeting;
• asking whether a person is free to talk now;
• acquaintance;
• presentation and discovery of interest;
• dealing with objections;
• closing with a desired action (with a deadline).

The speaker asserts that almost all the actions, save for the last two points, can be conducted by a bot quickly, reliably, and cheaply. Therefore, 90% of calls that are inefficient in call centers (for instance, a person is busy to talk or not interested in the offer) are removed from personnel responsibilities.

Tasks solved by voice bots

Considering voice bot possibilities more closely, one can determine several goals where the technology will be convenient.

Sales. From 70% to 90% of calls are ended in either a refusal or a request to call back, which is the waste of employees’ time. Thus, such a work can be shifted to a robot that will redirect people interested in the offer to the staff. In this case, there will be 10-30% of useful calls.

Surveys. A survey with closed questions requiring affirmative or negative answers can be easily automated. According to Aleksey Skripka, a common call center needs a month to complete this task, while a voice bot is able to implement this in a day. Bots can also process open questions, but it is a more complex pattern because the personnel will have to analyze the answers afterwards.

Notifications. Companies have been sending voice notifications to customers since the 1980s. Back then, phones replayed a certain, previously recorded audio track. The efficiency of similar calls in terms of conversion was only 1-2%, as people realized that it was a record and hung up. Talking to a modern bot, customers think that they communicate with a real employee.

Besides, a call is more profitable and successful than a text message. Firstly, three times as much information can be rendered using a voice at the cost of SMS. Secondly, a text message is paid regardless the result, while the value of voice bot services is estimated at the amount of notifications listened by users.

Phone support. A voice bot is able to replace answering machines, IVRs, and first line technical support asking standard questions. Bots can rapidly pick up the phone and answer a call. In particular, it can ask the name and reason for request, specify the way how a person is registered in the system, and then redirect the call to a particular specialist. It saves customers’ and call centers’ time.

Advertising. To prepare efficient advertising and product promotion, one should necessarily describe an appropriate offer and segment the base. Otherwise, calls will be poorly performing. The first step is segmentation: one creates a customer profile. An offer is adjusted to this client; specialists model a voice dialogue scenario (ranging from 10 to 20 response options) and record an audio track. The best possible bot is the one sounding like a human being so that customers have no idea they are talking to the machine.

How to make voice bot speech realistic

The productivity of automated calls depends on how realistic the bot sounds and how similar its speech and responses are to people’s.

“Now, the efficient bot should contain three aspects: the best technology in terms of recognition and determination of meaning; the best voice and scenario to make it as realistic as possible; and, surely, seamless implementation,” Aleksey Skripka stresses.

The speaker mentioned that EVE.calls tested almost all available recognition systems: Google, Nuance, Amazon, Yandex (speech technology center). As the result, it selected only Google and Nuance in terms of language recognition quality and file processing speed. Moreover, there are few systems able to process Russian and Ukrainian.

There is also a range of features related to voice announcement that should sounds like an ordinary talk rather than reading from a piece of paper. That is why EVE.calls invites professional actors to articulate certain speech scenarios.

Comparison of call center and voice bot expenses: case studies

Expenditures for calling organization made by a conventional call center include recruitment, personnel training, wages, office, taxes, CRM, telephony, motivation, and performance control.

In comparison, a bot calling service is provided on a turnkey basis. Besides, the system operates around-the-clock, customers are provided with CRM included into the price, assistance of manager, marketing specialist, and speaker, as well as with telephony, server, and speech recognition system.

Let’s review this in practice. A case study of hiring Uber driver implemented by two people cost the company $1500. The voice bot spent a single day for the same amount of calls (involving one team member to catch people interested in the offer). It cost the company $33 at the database of 2000 people.

A similar case study was the invitation to the Business Molodost event. Its database consisted of 8000 people. The company spent $70 for bot calls requiring a day. fterwards, two specialists worked with objections and closed the registration. If they did everything manually, it would take a month of work engaging six employees at expenditures of $4500.

Conclusion

Voice bot services can be used by any company requiring rapid calls through cold and hot databases. Not only will this method help to save time and money dozens of times, but it will also allow to relieve personnel’s operation. Using the automated call system, one can reduce a number of shifts and release staff from routine tasks. Instead of this, people will be in charge of communication only with the target audience interested in offers.