We already know successful cases of speech synthesis implementation in Ukrainian companies. But speech Analytics is still a little-known topic.
Speech Analytics in the contact center is the analysis of operators ‘ conversation records by keywords, queries and even intonations. That is tens of thousands of records a day. At the same time, the Quality Control Manager listens to about 32 calls a day. For one 3-minute conversation, they need to allocate 15 minutes of their time. Listen to the recording, set estimates, verify the data entered in the course of the conversation in CRM, fix the feedback to the supervisor for further work with the operator.
Quality control departments estimate up to 12% of calls. Speech Analytics estimates up to 95% of calls. Although it maybe 60%. This error is associated with interference on the line (the client is hard to hear on the street because of the wind), intonation, which determines the meaning of what is said, dialects, accents, etc.
Rogers in the US, by implementing speech Analytics, reduced the number of repeat customer calls by 25%. Issues were resolved at the first appeal in 80% of cases. In addition, revenue per call increased by more than 100%.
In Russia, speech Analytics was introduced for the first time in the National recovery service. If during a telephone conversation there are troubles – long pauses, quarrels, disputes – it immediately becomes known to the supervisor.
Among Russian banks, TransCreditBank was the first to install an automatic voice service. Its customers learn about the addresses of branches and ATMs without the participation of the contact center operator. For information, it is enough to call your region and city.
Speech Analytics helps to find out which of the operators incorrectly greeted, inaccurately answered the client. Nevertheless, the most important thing is that the system calculates what requests customers make and what they think about a particular offer. This is big data Analytics, known as data mining. The system calculates how consumers reacted to the product by analyzing their responses to the operators. Moreover, advanced systems will even determine the mood of the client at this moment. This option already allows you to identify annoying factors in the conversation.
Therefore, the Manager can see calls by type on the example of one of the speech analysis systems. Blue-standard calls. Yellow — problems with the quality of the employee. Red — a missed sale. Green-positive feedback. Purple is a client at risk.
For speech Analytics, the conversation recording is divided into 2 streams: the voice of the client and the voice of the operator. Each part is analyzed separately
Vendors say that it is worth implementing a speech Analytics system if there are more than 100 operators in the call center. However, it is necessary to estimate not quantity, and payback of this system, investments are big. Optimistic forecast of the return on investment is 1-2 years, while more realistically 3-5 years.
It will also take time to install and configure the system. This is about 2 weeks, plus 2 months for paperwork (description of the rules, legal aspects of voice processing, integration into other systems, etc.).
Speech Analytics systems know about the 10 most popular languages in the world. Among them – Russian, but not Ukrainian. This is a big drawback for Ukrainian companies.
Speech Analytics is unlikely to become a trend in the next year or even three. Its implementation requires large investments, and all its tasks are now solved by quality control and Analytics departments. But in the next ten years, the automation of these tasks and the processing of large amounts of requests will become inevitable.