PurposeEnabled by increased (“big”) data stocks and advanced (“machine learning”) analyses, the concept of human resource analytics (HRA) is expected to systematically improve decisions in human resource management (HRM). Since so far empirical evidence on this is, however, lacking, the authors' study examines which combinations of data and analyses are employed and which combinations deliver on the promise of improved decision quality.Design/methodology/approachTheoretically, the paper employs a neo-configurational approach for founding and conceptualizing HRA. Methodically, based on a sample of German organizations, two varieties (crisp set and multi-value) of qualitative comparative analysis (QCA) are employed to identify combinations of data and analyses sufficient and necessary for HRA success.FindingsThe authors' study identifies existing configurations of data and analyses in HRM and uncovers which of these configurations cause improved decision quality. By evidencing that and which combinations of data and analyses conjuncturally cause decision quality, the authors' study provides a first confirmation of HRA success.Research limitations/implicationsMajor limitations refer to the cross-sectional and national sample and the usage of subjective measures. Major implications are the suitability of neo-configurational approaches for future research on HRA, while deeper conceptualizing and researching both the characteristics and outcomes of HRA constitutes a core future task.Originality/valueThe authors' paper employs an innovative theoretical-methodical approach to explain and analyze conditions that conjuncturally cause decision quality therewith offering much needed empirical evidence on HRA success.