We've had 329 items through our submission form since Christmas 2008. Of these, our Repositories Assistant, Marie has submitted 221 records through that form. Which means that we've had 108 articles sent through administrators or by authors themselves, through our form.
Most of the articles that Marie has submitted have come from e-mails sent by authors to her, again most of which were generated through personal invitations as she scanned their web pages before writing to authors.
Some of those personal invitations will have been based on Zetoc alerts. Out of 124 alert-based emails sent to authors since Jan 2008 we have had 29 articles deposited, which makes a 23% success rate for those e-mail invitations. At first, very few alert-based requests were made, but this year we have sent out about ten a month, as a rough average. We are sent alerts by helpful subject librarians who monitor lists of names for a single department, and are able to quickly delete those which don't corresond to their department's subject. For this reason, one person can only monitor alerts for a single department so only a handful of departments' authors are being monitored, and authors with common names cannot be monitored at all. Subject librarians investigate alerts for items which appear to be journal articles in the correct area of interest for the Warwick author of that name, and forward alerts of Warwick authors' articles to Marie.
Marie's processing of the alert involves checking whether we've already got the article in WRAP (it has happened once or twice!) and then looking at the publisher's policy, to see if she can just put the final version into the repository and let the author know, rather than writing to the author to ask them for a post-print. This check also avoids asking for an article that cannot be put into the repository in full text.
Looking back through our alerts, I can see that we have written to some authors more than once: one author has written five articles, and we have written to him every time. Although our statistics show that that author has not deposited as a result of those alerts, I do know that individual since he followed our first alert invitation up with an enquiry, which resulted in deposits of other articles to WRAP. However, it might be worth our while considering whether to continue monitoring alerts for authors who never respond to our invitations (particularly those with the most common names, who generate a larger number of "non-Warwick" alerts which must be gone through every morning), or at least following up our e-mail invitations with a phone call. We have never chased anyone about an item we have invited deposit of. At least one other author has sent us other articles than the one we requested through an alert, from amongst his back catalogue of published work.
We are also looking at using other alerting services, and are currently investigating PubMed with its author affiliation searching.
Our Zetoc alert based requests would seem to be the most successful approach to authors to date, although we are now in a position to use other advocacy tactics, since the repository is well established and growing in size. We will, of course, continue to monitor the effectiveness of all our advocacy tactics...