The current offer IVR menu options in a pre-static tree. This menu, apart from variables that are going to follow to identify, is presented, but always in the same manner. Determined and decisive factors (such as call history, time slot and the time of delivery) can help improve the quality of service provided and the number of requests handled. Developing appropriate variables considered, it is possible to predict the major and most recurrent requests of customers, enabling the dynamic reorganization of the tree of I-IVR options.

We designed an algorithm to estimate applicable to Interactive Voice Response in order to make this tool capable of making predictions "near real time" based on requests from users, thus giving birth to a new generation of intelligent answering machines:
Intelligent – Interactive Voice Response.

Technology deployment

The implementation of the algorithm estimates developed by us can be through any programming language and any operating system. Can be integrated into existing applications or add them as external (eg CRM or CTI).

Details on the implementation

The Intelligent IVR-is composed of a service system and a web interface for administration of the functionality offered by the service. The I-IVR once configured can operate "stand alone", managing, with complete autonomy, the entire architecture dedicated to the service response.
The system can be extended on request with CRM functionality. It can also be integrated on existing platforms (eg Genesys), or as software / firmware on some native types of devices (eg Avaya).

Forecasting and management

Through the I-IVR, and through the detection and analysis of past trends, the dynamic reorganization of the menu options will be handled through rules (to be defined by mutual agreement) as follows:

• achievement of certain thresholds (temporal or numerical, to the single minute or single call to make, in fact, practically constant adaptation of the menu offered);
• template preloaded software;
• manually (on site or remotely) by means of a simple "drag & drop".

Whatever the mode in use, the update of the menu is done "on the fly", without arrests or delivery of the service machine, so if you realized, during a phone call, the condition of change the I-IVR, the user in line would continue to use the active menu on the start of his call for the next while, the options are presented according to the new requirements, together with old ones. The options constant, which usually do not exceed 2-3% of the total, will not be subject to relocation dynamics (for example, marketing messaging, an option trader and others).

Model Forecast

The forecast model can apply to different parameters used for prediction.
The relationship between minimum time of sampling parameters and time of application of the forecast, as well as the simultaneous existence of multiple active trees is determined by the characteristics of the hardware detection systems and data processing and application of forecasts.
Minimally supported in the hardware, the algorithm is able to create and successfully manage a high diversity of trees and posts, offering the highest standard of quality of services available on the market.
It's possible to implement the system starting from simple forms through to statistics of the most sophisticated neural network systems.

Applicability

The architecture can be applied both locally, ie the parameters of detection in a given I-IVR, both distributed (unlike the traditional systems), ie the parameters of detection of a "pool" of I -IVR, providing for the same service, spread over several sites.

Benefits

- Reduction of unnecessary time in the navigation menu options

- Reduce the average time of call duration

- Increase the catchment area servibile

- Reduction of costs for the management of free services to the user

- Changing the structure of the nodes in an autonomous and without suspension of    operation

Brochure in pdf format



To view pdf documents you can download Adobe Acrobat Reader