The main objective of this work was to group altmetric indicators according to their relationships and detect disciplinary differences with regard to altmetric impact in a set of 3793 research articles published in 2013. Three of the most representative altmetric providers (Altmetric, PlumX and Crossref Event Data) and Scopus were used to extract information about these publications and their metrics. Principal component analysis was used to summarize the information on these metrics and detect groups of indicators. The results show that these metrics can be grouped into three components: social media, gathering metrics from social networks and online media; usage, including metrics on downloads and views; and citations and saves, grouping metrics related to research impact and saves in bookmarking sites. With regard to disciplinary differences, articles in the General category attract more attention from social media, Social Sciences articles have higher usage than Physical Sciences, and General articles are more cited and saved than Health Sciences and Social Sciences articles.