Data Driven Usability: A Case for Adaptive Interfaces in Voice Based Menu Systems

Interactive Voice Response (IVR) systems provide access to information over the phone by responding to a pre-defined menu either through key presses or voice commands. Despite being in use for a long period, IVR systems are still considered time-consuming and frustrating to use. In this work, we show that a significant portion of the call duration goes in selecting the correct menu option. Since the menu options are presented sequentially, more time is required to access menu options appearing later in the sequence, therefore, to reduce this, relevant menu options must appear early in the sequence. In this paper, we present our data driven prediction algorithms for adaptive rearrangement of menu options so that the relevant options appear early. We also show that adaptive approaches to decide the menu structure outperform existing static menu based IVR system. We have designed, deployed, and evaluated our schemes in the real world study.