The last decades witnessed the rise of innovative development, based on scientific inquiry, as a priority of national policies in numerous countries (including Russia). Growing attention to the research activities is accompanied by the search for better methods to assess scientific productivity of individuals, institutions and entire professional communities. The current article focuses on the bottlenecks in this process as well as possible ways for their overcoming building upon recent record of the United States. This country represents a particular significant case for study as it maintains global leadership in both fundamental and applied research. The article highlights the necessity to differentiate between two levels of assessment: the one dealing with immediate scientific outcomes and the other reflecting broader social consequences of research. Both state and society are more interested in the latter; however, they are dependent on the whole innovation system of the state, rather than on scholars alone. The study of scientific productivity relies on bibliometric methods and expert evaluation. Despite the surge of interest towards quantitative assessments, their uncritical application could lead to the significant distortions of the real impact of individual scholars in professional community. They could play useful supplementary role, but they do not provide solutions for making certain managerial decisions. Aggregated data is not specifically useful for forecasting productivity of individual scholars. In the assessment of broader social contribution of scientific research in the U.S., there is a growing appreciation of multidimensional analysis, relying on Big Data methods. Currently, it is underdeveloped; however, the demand from American federal authorities stimulates rapid development of evaluation instruments for better targeting of policy.