RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2
Project

JSC Aeroflot implements the solution of speech recognition and routing in the contact center

Customers: Aeroflot

Moscow; Transport

Product: Nuance Call Steering

Project date: 2010/03  - 2011/12
Project's budget: 94 million. руб.
Number of licenses: 130

Content

The project team from the customerIntegrator Consultant
Bogdanov Kirill Igorevich (directors of IT)
Yurchenko Vitaly Eduardovich (Director of technology development)
Ushakov Alexey Aleksandrovich (Technical director)

Nuance Communications announced on December 19, 2012 implementing solution of routing of incoming calls on the basis of speech recognition of Nuance Call Steering in contact center of Aeroflot company.

Previously

Before start of the Call Steering "Aeroflot" platform used the distributed contact center (CC) consisting of three physical platforms from two independent operators of services outsourcing CC. As point of entry for calls of clients served the interactive voice answering machine IVR with distribution of calls by set in a tone mode.

Problems:

  • lack of incoming lines;
  • progressive tense of waiting of the reply of operator;
  • low controllability (difficulty with introduction of operational changes in the scenario of call processing);
  • high cost of operation;
  • lack of uniform quality control of call processing.


Developers of the project were faced by problems of automation of work of CC by reduction of a share of work of operators and transfer of a considerable part of customer appeals to self-service in the automatic mode, cost reduction on the maintenance of contact center, maintenance of image of Aeroflot - the innovative company.

Choice of the supplier

2010

To solve assigned tasks, Aeroflot announced competition on the best solution on automation of call processing designed to simplify routing of calls and to automate providing information which is often requested by passengers. The solution had to consider a possibility of flexible start of new interactive voice applications. Competitive requirements were formulated jointly by IT department and the commercial block. As a result of the held competition Aeroflot stopped on the option offered by Voxcom company – implementation of the voice Nuance Call Steering platform. Alternative offers from other suppliers assumed use of normal interactive voice answering machines (IVR) with functions of tone extension dialing of number for distribution of calls by queues.

Automation of work of contact center concerned several applications:

  • management of incoming calls (distribution by groups of operators or services of autoinforming);
  • prearmoring (the reference on runs and existence of tickets without commission of direct armoring and payment of tickets);
  • the personalized outgoing voice notification of passengers about changes in the schedule of the flights performed from Moscow (change of terminals, delays and cancellings of runs, etc.);
  • outgoing informing on stocks of the company (in particular, change of conditions of the Aeroflot Bonus program in April, 2011).

Features of the project consisted in need of integration of the Call Steering platform with the global Sabre booking system and additional internal data sources of JSC Aeroflot. One more complexity was creation of the reporting for tracking of key indicators of system operation in real time.

The task was complicated by what at the same time the Sabre booking system was transferred to architecture of web services which at the time of project implementation was still insufficiently well documented. Also CTI-integration of the Call Steering platform with software of outsourcing contact center was required.

Nuance Call Steering

Often at the appeal to contact centers clients are forced to make the way through the whole labyrinth of menu items unfamiliar to them. Considerable number of calling refuse to understand logic of the menu of interactive voice answering machines (IVR) and at the first opportunity try to unite to any of operators. Meanwhile the applications of self-service which are available for the company can stay idle because clients do not find them and, respectively, cannot use them...

Nuance Call Steering allows to connect clients to that service or those operators who are necessary to them quickly. The recognition of the natural speech based on statistical models for processing of incoming requests allows to use the solution. Call Steering delivers requests of subscribers in the necessary point for smaller quantity of steps in comparison with other technologies of hands-off processing of calls.

Serious restrictions as they process only a certain set of answers (phrases) are inherent in many systems of speech recognition. They cannot "study" and expect what the client can tell. Unlike such systems, in the solution

Nuance Call Steering are used such developments in the field of speech recognition as Nuance Recognizer which supports statistical semantic modeling of language.

These technologies help to create extremely effective applications of voice self-service that leads not only to significant improvement of quality of service, but also to decrease in operating expenses of the companies.

The solution has the developed tools of monitoring and analytics. The possibility of the analysis of collected data on request processing, regular tracking of quality of work with assessment of accuracy of recognition and identification of problem points of routing allows to reveal and fix timely problems with speech recognition and routing of calls. In turn, such functionality constantly provides the shortest way of processing of client requests.

Use of effective recognition of the natural speech for management of incoming calls saves money thanks to automation of process of service of requests and reduction of a share of wrong connections. Simplification of the procedure of provision of services, in turn, increases customer loyalty level.

Process of implementation and setup of a system took about six months. According to the results of project implementation economy of the budget of airline on the maintenance of contact center made over 31 million rubles in a year. 2.1 million calls were during this time processed. During the separate periods peak load on a system reached 150 thousand calls a day. Analyzing requests of clients, a system is capable to route telephone calls in more than 30 different directions. Now instead of three distributed platforms of contact center only one is used on 80 operator places.

"Serious restrictions as they process only a certain set of answers (phrases) are inherent in many systems of speech recognition, they cannot 'study' and expect what the client can tell — Scott noted Vikveyr, the regional vice president of Nuance. — Unlike such systems in the solution Call Steering the advanced developments in the field of speech recognition are used, in particular Nuance Recognizer which is the most successful system of speech recognition in the world and supports statistical semantic modeling of language".

Project implementation provided increase in processing speed of each call and reduction of number of wrong connections. In connection with decrease in quantity of the unprocessed calls, for last year the number of the calls accepted by contact center of Aeroflot increased by one and a half times. Using the Call Steering platform in a year since system launch in operation more than 2.1 million calls are processed.

Economy of the budget of JSC Aeroflot on the maintenance of contact center in a year made more than 31 million rubles. Instead of 3 distributed platforms of contact center one is used (80 operator places). Now more than 60% of references on runs are issued to clients of JSC Aeroflot in completely automatic mode.

After setup of a system the accuracy of recognition of requests in the separate directions (general information, new armoring, the Aeroflot Bonus program) reaches 96.5%. On 44% of requests clients receive the automatic reference and switch to operators for armoring implementation. The notification of passengers about changes of the schedule of runs is automated for 99.8%.