Automatic Speech Analytics as a Technology of Today

Effectively functioning contact center is an important condition for company competitive advantage in growing competition on the modern service market. Such contact center is characterized by excellent customer service and low current expenses. Unfortunately, the role of contact center is often underestimated in the company. Every day agents receive a huge flow of information about quality problems, claims, strategies of the competitors, new laws, success of promotion actions, ideas about new services and products. Large companies spend hundreds of thousands dollars for market researches not mentioning the fact that they already have a large number of market information. And this information is free of charge and comes to contact center along with calls.

Nowadays we live in unique times for technological re-equipment of contact centers. Several years ago speech recognition in call centers was considered as a technology of the future, and today the system of speech analytics is an operating tool for invaluable information search. Such combination of an assistant, a controller, an analyst and an expert in one product permits to solve problems quickly and effectively with the help of complex tracking of numerous parameters: control of telephone data, call transfers and on-holds, initiating of call cancellation, number of repeated calls; correlation of agent’s and customer’s speech duration, number of interjections, detecting of simultaneous speech, tracking of undesirable phrases, monitoring of customer’s emotional state and etc.

There are three areas, where speech analytics system raises the performance of contact center and of the whole company:
• monitoring of agents who do not cope with certain situations during the dialogue because of insufficient professionalism or weakened motivation. It allows to define, what kind of coaching is necessary for the agents;
• deep understanding of customers’ requirements and preferences;
• business processes optimization.

Developed automatic analysis system helps to optimize main objective of a contact center in a short time period – to provide excellent service to customers. Optimization is implemented by KPI improvement. As an example we consider four cases of contact center overall performance increase according to the following key performance indicators: AHT (average handling time), FCR (first call resolution), CS (customer satisfaction), QM (quality management).

Conversations can be elongated reasonably and unreasonably. It is important to provide analysis for performance decreasing in cases of unreasonably long conversations. It is recommended to control the following factors: Percentage of customer’s speech. High percent of this parameter testifies the calls from aggressive or talkative customers. If the percent of customer’s speech is low it indicates that the customer is passive (silent). Correct and prompt coaching devoted to coping with different types of customers will help to reduce dialogue duration. Much time spent in on-hold mode, long silence and a large number of calls put in mute mode show lack of knowledge and competence and insufficient motivation of agents, inconvenience of business processes. System of automatic speech analytics helps to find proper and well-timed solution of the problems dealt with AHT improvement. Besides, speech analytics increases efficiency of quality managers, allowing them to use not less than 75% of working hours for coaching and agent leading.


Experience of unique speech analytics technology implementation proves its indispensability in the contact centers aspiring:
• to provide high-level control of main key performance indicators in a contact center;
• to build and to support motivation systems for contact center staff on the basis of objective measured indicators;
• to create a staff of competent employees;
• to reveal situations which demand coaching and special quality monitoring;
• to expose the influence of communication regulations on the quality of provided services;
• to increase percent of service satisfaction.

Listening to the chains of soundtracks which are repeated calls from the same customer allows to reveal the reasons of low FCR and to optimize service level of certain agents or business processes. A customer who calls repeatedly and tells agents about his multi-call experience seeks to draw special attention to his demand. Conversations with customers’ speech consisting of phrases, which indicate repeated calls must be in the first priority of problem solution since such calls are potential claims. Automatic tracking of emotional state and disloyalty detection allows to react to the claims which happen as a result of low service level. Insufficient attention to such soundtracks can lead to client base outflow.
Clients show discontent and use emotions and/or claim phrases. Unfortunately, not all claims are being transferred to the second level of service. Soundtracks with high percentage of customer’s speech often signal that it’s a claim call. The accuracy of claim control can be increased by tracking this factor along with customer’s emotional state and presence of conflict phrases. Besides, speech analytics system can display (in automatic mode) agents who procrastinate in picking up the phone, thereby causing extra claims. Call transfer monitoring helps to find the dependence of emotional customer’s state upon quantity of transfers.
Quality management allows to reveal cases of inadmissible agent service, to evaluate interactions, to provide control and to contribute to better agent work. It leads to efficiency rising in a call center. Undesirable phrases are the words which are accepted by customers’ imagination as negative. A customer has an impression that his demands are not important, unresolvable, agents do not possess professionalism, and the company intentionally deceives customers. Incorrect behavior is unacceptable for agents. It is prohibited to let customers know that they are guilty. Agents cannot mentor customers in a rude manner. The agents who use undesirable phrases provoke conflict. Agents, who interrupt customers, show weak emotional competence. Inadmissible case of service takes place when a call is dumped by an agent. It is important to trace such situations in call centers and provide necessary work for quality service improvement.