All new visitors
I had another little play with Google Analytics this morning, just because I happened to be signed in to Google (looking at the Australian repository managers Google group). Today I explored the "Advanced Segments", available in Beta.
I compared "New visitors" to "All visits" in the last month, and the trend is remarkably the same pattern, just slightly lower. Very few people come back to WRAP after having visited once: it is not a site that people wish to visit and explore. They dip in for the content that they want, and they don't necessarily come back.
Our "Bounce rate" is also just over 71% and this also confirms the pattern of visitors going in and then back out again. Their landing page is therefore the one they want, and they follow our link to the published item (which is what we want). Or else they were looking for something else and they go back to their search results. They might have read our abstract and learnt enough about the paper to know that they want something else. Or they might not have wanted an academic paper at all. We can't actually tell where visitors go when they leave WRAP.
Our highest bounce rate amongst the top content pages is our Information page (80%), which is also to be expected because the main way we use that page is to link to content on the Library's website, which is far more easy for us to edit and keep up to date. The next highest is for one of the papers, which is 72%, and so far as I can tell at a glance, this is typical of the bounce rates of our papers. Bounce rate brings in an interesting different dynamic for when I write to our top content authors... one with a high number of visitors who also remain in WRAP (about 54%) is also one which for we do not make full text available. I looked at the Navigation summary for this page, to try to get a feel for where the visitors to that paper were coming from and going to. 25% had come from a previous page in WRAP, and 29% went on to another page in WRAP. Mostly, people seem to have been clicking to see the pdf. But the locks are all working well, so people must be just testing them. The paper will be released from embargo at the end of this month. Other visitors did go on to view another paper by the same author, and some of them were clicking on our subject heading. I rather suspect that this anomaly is the result of my using this author's RSS feed on my example page, to show people how easy it is to incorporate WRAP results into a Warwick web page! I chose this author's content because it was already popular, but I will keep an eye on top contents' bounce rates in future.
The general trend of visitors dipping in and out of WRAP rather confirms my impression that most people will find repository content when searching elsewhere, rather than searching in the repository itself. This may change as our body of content grows. But for now, even though we're pleased to be approaching 1000 items, our content is pretty disparate across the subjects and as such, not the most useful of collections to researchers.
Referral traffic on WRAP is slightly higher than direct traffic, and both are bouncing around that zero line but again, it is search traffic that follows the overall pattern of visitors. Perhaps academics will get used to sharing WRAP URLs in the future, and we will get more referrals.
The pattern of visitors to WRAP pretty much confirms our dependency on Google and its ilk to bring us visitors. Which is OK, but a slide that I saw yesterday from Tom Abbott, who co-ordinates all Warwick's iTunesU output demonstrated the difference that being a destination site (or promoted on one) can make. The usage graph he showed, for a file on Warwick's own website, which was then published on iTunesU showed a vast increase. Visitor numbers for the podcast on Shakespeare's portrait seem to have been astronomical as a result of Apple promoting it amongst their collection (millions of visits rather than hundreds, as with WRAP!). iTunesU is a destination site. Repositories like WRAP are not, but perhaps repository cross-searching sites could be. That's a part of the reason for making our metadata such high quality and harvestable.
Looking at the highest referring sites to WRAP gives some clues as to who might become more important. The Index to Theses (http://www.theses.com/) is likely to be a destination site for some. They link to us from: http://www.theses.com/idx/registered_users/etd/96.aspbut as yet don't link to our metadata records from theirs. Fair enough, 'cos we've been a bit slow off the mark with the theses. But we're adding them now, and as we come to add more theses, this might be a source of more visitors: they do have an example of a Cranfield thesis with a link to the full text in the repository from their record. Although I didn't think that the link was all that obvious: it's in the left hand margin, rather than in the text of the record describing the item. So many academic resources seem so un-user-friendly!
Other referrers to WRAP include bing.com, which is actually a UK specialist search engine (and has recently become the second highest search engine source of visitors to WRAP), google itself (in various guises), Warwick's own research information system (whose records link to ours) and the NHS Evidence Health Information Resources website. I'm not sure why Bing and the Google guises reported in this section are not included in the search engine hits... but at any rate, I consider them search engines!
OpenDoar and ROAR brought us three visits each... and I think that at least one of those visits would have been me, checking our record with them! Repositories and repository cross-searching tools are not destination sites in their own right yet (or at least WRAP isn't one: perhaps larger repositories see a different pattern). Repository cross-searching sites are not bringing us visitors. Search engines are...
I don't know of anyone who does research by visiting a repository cross-searching tool. I don't believe that they're even promoted by academic librarians in their liaison with departments. OpenDOAR and ROAR seem to me to be more tools for those in the repository and information management sector. My personal favourite cross-searching tool is OAIster, but I haven't investigated them all thoroughly, and I think it's a side of repositories that deserves further investigation. Now that we've got content in them, with high quality metadata, how do we ensure that people will find it in ways convenient to them? Google is great but it only goes so far...
Once we have tools that we might recommend people to use, to find repository content, that's a whole other advocacy/information literacy journey we would need to go on, persuading researchers of how to find and use repository records. Our Psychology librarian has made a start: her tutorial for undergraduates, which she created in collaboration with tutors in that department, recommends that students might like to look at their tutors' papers in WRAP, so that they can learn about their areas of expertise and interest and get ideas for their projects. But there is a long way to go!