Scientific research produces new knowledge, technologies, and clinical treatments that can lead to enormous returns. Often, the path from basic research to new paradigms and direct impact on society takes time. Precise quantification of scientific output in the short-term is not an easy task but is critical for evaluating scientists, laboratories, departments, and institutions. While there have been attempts to quantifying scientific output, we argue that current methods are not ideal and suffer from solvable difficulties. Here we propose criteria that a metric should have to be considered a good index of scientific output. Specifically, we argue that such an index should be quantitative, based on robust data, rapidly updated and retrospective, presented with confidence intervals, normalized by number of contributors, career stage and discipline, impractical to manipulate, and focused on quality over quantity. Such an index should be validated through empirical testing. The purpose of quantitatively evaluating scientific output is not to replace careful, rigorous review by experts but rather to complement those efforts. Because it has the potential to greatly influence the efficiency of scientific research, we have a duty to reflect upon and implement novel and rigorous ways of evaluating scientific output. The criteria proposed here provide initial steps toward the systematic development and validation of a metric to evaluate scientific output.