Online social networks act as good mediums for communication but are also becoming popular for targeting the social needs of their users. Mainstream social networks are still unable to incorporate lowliterate users into their user-base as their interfaces are on the web or in Short Message Service (SMS) format, while low-literate people constitute a major portion of world’s population. Speech-based networks (SBNs) overcome these limitations by providing a simple speech-based interface for users. In this work, we present a systematic analysis of SBNs designed specifically for low-literate users with a focus on identification of influential users. The task of finding influential users has not been studied for SBNs. Furthermore, knowledge of influential users can help optimize the operation of SBNs that typically run in low-income regions with low budgets. We demonstrate how a SBN is formed from call data records and define its key features or characteristics. We then propose a feature-based method for influence ranking in a SBN. Existing methods for influence maximization in social networks are not directly applicable to SBNs. Hence, we present a method for calculating influence probabilities between users in the network, enabling the application of the greedy algorithm and degree discount heuristic for influence maximization and computation of betweeness centrality in a SBN. We evaluate our methodology on data from a real-world SBN called Polly. We compare the results with those from existing methods and show that our methods are both effective and time-efficient for use in SBNs.