Thomson Reuters uses altmetrics: usage counts for articles indexed in the Web of Science

Abstract

Since September 2015, Web of Science has displayed an indicator of a non-bibliometric nature. The new indicator, called "usage count", measures the number of times the full text of a document has been accessed, and the number of times the bibliographic reference of the document has been exported to a reference manager. The same indicator is computed for two different time frames: one is a rolling window that measures the number of times the document has been accessed in the last 180 days and the other measures all instances of access since 2013. In order to decipher the meaning of this new indicator, apart from reflecting on the actions we expect to measure, an exploratory study was conducted where the new indicators were compared to the traditional citation counts, by pulling two samples of documents: the most cited documents published in 2015 (so far), and the most used ones. The greatest difference found between the two samples indicates the two indicators reflect different types of impact. After all, citations and usage of documents are actions so different in nature that their counts can only offer divergent results. Moreover, a very low correlation was found between these usage counts and the most common altmetrics. However, given the size and bias of the sample (highly cited and highly used documents), the results should be considered with caution. Nonetheless, we can be sure that usage counts enrich the set of bibliometric indicators offered by Web of Science, and that they will provide new insights about previously disregarded documents. Empirical data available from: http://dx.doi.org/10.6084/m9.figshare.1564716