Without many words
How can banks monitor compliance with amendments to Law No. 39-FZ using speech recognition technology and speech analytics
In December 2018, amendments to Law No. 397-FZ entered into force, designed to tighten control over the activities of financial advisors in Russia. Innovations of the Central Bank - an attempt to create a domestic system of control over the activities of financial advisers, similar to the American FINRA. In other words, the Central Bank of the Russian Federation is going to monitor the quality and liquidity of the investments offered to customers. The Law covers all persons and organizations calling themselves investment advisers and advising clients on financial investments.
The initiative of the Central Bank of the Russian Federation is favorable for clients and is quite justified - earlier the procedure for the activity of investment advisers was not regulated at all. Any person, regardless of experience or qualifications, could call himself a financial consultant. Offers of low-quality or extremely disadvantageous financial products often brought people problems.
For customers, the benefits of CB innovations are obvious - on the one hand, a significant increase in the quality of financial advice and lower risks, on the other hand, the outcome from the consulting market of fraudsters or non-professionals. But what does this mean for banks?
"Secret" regulation
In 2018, the Central Bank decided that traditional audit tools to monitor compliance with Law No. 397-FZ would not be enough, new ones were needed, so the idea came up to look for violations using "The Mystery Shopper" method. It is assumed that central bankers in charge of banking will call branches and call centers and introduce themselves as potential customers. So the regulator plans to look for violations in financial organizations.
According to Ivan Tertychny, a partner of Tertychny Agabalyan, the Central Bank wants to obtain the status of a subject of operational-search activity. In this case, the regulator will receive the right to conduct test purchases. Consequently, calls by the Central Bank staff will be considered a supervisory measure.
To monitor compliance with amendments to Law No. 39-FZ, banks will need to adopt local regulations
By themselves, these data will not become the basis for the application of supervisory measures, but in the future the collected materials can be transferred to the Investigative Committee or the Ministry of Internal Affairs. In addition, the regulator will be able to send unscheduled inspections to banks, which will confirm and record violations in the inspection act.
To monitor compliance with amendments to Law No. 39-FZ, banks will need to adopt local regulations. In addition, you will have to control the communication process of the organization’s employees when interacting with customers in terms of financial advice.
According to the provisions of the Law, the bank is obliged to provide the client with information about all investment products, to notify that the consultation is not an individual recommendation, and the previous results of profitability do not determine future profitability. It is necessary to build an offer of any investment product strictly on the basis of the identified investment profile of the client. Deviations from the provisions of the Law bear the risk of financial loss for the bank and require prompt response.
What to do?
In the future, any consultant who communicates with a potential client may fall into such a "mystery shopper." For a banking organization, these are huge risks. It is impossible to predict when, to whom, and on what matter an employee of the regulator will call. In addition, it is unlikely that he can be identified in any way. To stand behind all consultants and to control every word, of course, there is no physical opportunity.
Of course, you can spend a lot of money on training employees so that their speech complies with the requirements of the law, but this will not save from simple human forgetfulness or inattention, which can be an argument for taking measures.
An alternative method of regulating employees will be a more reliable solution in this situation. The level of development of digital technology allows delegating control over the speech of employees to algorithms.
Telephone calls
It is no secret that the vast majority of calls made by bank employees are recorded. But there are hardly any resources to listen to hundreds of hours of talk about compliance with the law. Records are usually used to resolve conflicts or when complaints are received from the client. However, telephone conversations are a source of valuable information, which, when used correctly, can not only solve the issue of compliance by consultants with the law, but also increase the efficiency of employees.
The full potential of information obtained from records is realized using digital technologies, in particular through speech recognition and subsequent analytics.
You can spend a lot of money on training employees so that their speech meets the requirements of the law, but this will not save from simple human forgetfulness or inattention
At the first stage, the call recording passes through an algorithm that decrypts the audio. Speech recognition is a fairly mature technology, the development of which the leading IT specialists around the world have been engaged in over the past few years. Recognition accuracy is close to 100%, this allows you to accurately transfer the contents of the conversation to the speech analytics system.
The second stage is the analysis of the information received. The intelligent system checks the employee’s speech for compliance with the script and the presence or absence of certain information blocks. The system works as follows: first, all possible formulations are introduced into it to mention a particular topic, thanks to various options, the employee does not have to literally speak the script, different ways of developing the conversation appear. After the word sets are formed, the logic of the script is checked - the dialogue sequence. The scripts themselves can be either cyclic or branched. When passing through each block, it is shown with what words the employee did it. Missed blocks are marked separately. Analytical systems also collect data on the frequency of use of certain words. By and large, analytics can be customized to any request and tracking any elements in the speech of employees. Statistical data for each employee can be found in the table, which is updated when new data is received.
Live Communication
If with phone calls everything is more or less clear (conversations are recorded), then a personal meeting with a client potentially looks much more dangerous. Recording a person speech without his consent is illegal. However, during a personal meeting of the client, no one warns that the conversation is being recorded, this can cause distrust, reduce loyalty and the desire to cooperate with the bank.
In this situation, a key role is played by the level of development of modern speech recognition algorithms. There is a technological solution based on a neural network that can cut unwanted elements from recorded speech. In this case, not only extraneous sounds are deleted, but also client’s replicas - they are simply absent in the record.
The solution itself is a voice recorder badge that does not interfere with either the employee or the client. The device records all conversations in a day, and then transfers them to the same analytical system.
Both solutions allow you to tighten control over employees who are in direct contact with customers, while not creating the effect of the presence of a "big brother." Nobody listens to conversations - the algorithm checks them.
Among other things, such systems become a good alternative to expensive training and help to create a self-education culture among employees. At the end of the working day, each manager is sent an individual summary report, automatically generated on the basis of all his conversations. Agree, it is much easier to develop your skills when there is visual evidence of weaknesses and strengths.
Speech technology, like many other digital innovations, is primarily designed to offload people from routine tasks. The presented solutions on average reduce the time spent by the head on personnel control by 2.4 times. In addition, employees who have access to reports on their own speech, on average, are 10 times less likely to use parasitic words in conversations and 2 times more accurately convey key messages from the script. In addition to all that has been said, speech technologies can increase sales of targeted products by up to 36%.
