Turning information quality into firm performance in the big data economy

Purpose: Big data analytics (BDA) increasingly provide value to firms for robust decision making and solving business problems. The purpose of this paper is to explore information quality dynamics in big data environment linking business value, user satisfaction and firm performance. Design/methodology/approach: Drawing on the appraisal-emotional response-coping framework, the authors propose a theory on information quality dynamics that helps in achieving business value, user satisfaction and firm performance with big data strategy and implementation. Information quality from BDA is conceptualized as the antecedent to the emotional response (e.g. value and satisfaction) and coping (performance). Proposed information quality dynamics are tested using data collected from 302 business analysts across various organizations in France and the USA. Findings: The findings suggest that information quality in BDA reflects four significant dimensions: completeness, currency, format and accuracy. The overall information quality has significant, positive impact on firm performance which is mediated by business value (e.g. transactional, strategic and transformational) and user satisfaction. Research limitations/implications: On the one hand, this paper shows how to operationalize information quality, business value, satisfaction and firm performance in BDA using PLS-SEM. On the other hand, it proposes an REBUS-PLS algorithm to automatically detect three groups of users sharing the same behaviors when determining the information quality perceptions of BDA. Practical implications: The study offers a set of determinants for information quality and business value in BDA projects, in order to support managers in their decision to enhance user satisfaction and firm performance. Originality/value: The paper extends big data literature by offering an appraisal-emotional response-coping framework that is well fitted for information quality modeling on firm performance. The methodological novelty lies in embracing REBUS-PLS to handle unobserved heterogeneity in the sample