The impact agenda in many countries has led to increased attempts to assess the societal impacts of research. Altmetrics, webometrics, and other alternative indicators have been proposed to support this task, and many journal articles have been written that exploit alternative indicators to investigate societal impacts. Nevertheless, methodological studies of many of these indicators have revealed that extreme care must be taken with gathering, aggregating, and interpreting them. This article gives an overview of current alternative indicators, summarizes empirical research, and reports a series of common problems and mistakes to avoid when using them. The main issues are: selecting indicators to match goals; aggregating them in a way sensitive to field and publication year differences; largely avoiding them in formal evaluations; understanding that they reflect a biased fraction of the activity of interest; and understanding the type of impact reflected rather than interpreting them at face value.