I’ve outlined a bit of how read posts on the web can send notifications to journal articles to allow them to better track traffic. Similar to use cases I’ve outlined for podcasts which have some large aggregate download data, but absolutely no actual “I listened to this particular episode” data, explicit read webmentions for journal articles could be a boon to these journals as well as to the greater research enterprise.
Separately but similarly, it would be nice if journals could take advantage of annotation platforms like Hypothes.is (especially if they sent webmentions to the canonical links or DOIs for .pdfs) to get a better idea of how closely, or not, academics are reading and annotating their works.