Decision Making in Enterprise Crowdsourcing Services

Abstract

Enterprises are increasingly employing crowdsourcing to engage employees and public as part of their business processes, given a promising, low cost, access to scalable workforce online. Common examples include harnessing of crowd expertise for enterprise knowledge discovery, software development, product support and innovation. Crowdsourcing tasks vary in their complexity, required level of business support and investment, and most importantly the quality of outcome. As such, not every step in a business process can successfully lend itself to crowdsourcing. In this paper, we present a decision-making and execution service, called CrowdArb, operating on crowdsourcing tasks in the large global enterprise. The system employs decision theoretic methodology to assess whether to crowdsource or not a selected step of the knowledge discovery process. The system addresses the challenges of trade-off between the quality and time of the crowdsourcing responses, as well as the trade-off between the cost of crowdsourcing experts and time required to complete the entire campaign. We present evaluation results from simulations of CrowdArb in enterprise crowdsourcing campaign that engaged over 560 client representatives to obtain actionable insights. We discuss how proposed solution addresses the opportunity to close the gap of semi-automated task coordination in crowdsourcing environments.