Usage Statistics

There are many reasons why an institution will have a repository. One of the main drivers of repository development has been 'open access' which places the research outputs of an institution into the public domain. Usage statistics can be used to show how many visitors are downloading the items held in the repository. This information can be used in a variety of ways, including:

• Tracking visitor numbers with the aim of increasing them over time

• Quantitatively analysing the popularity of items against each other

• Ensuring publicity campaigns are working

• Tracking inbound links to items

• Encourage further repository deposits by proving the usefulness of current items

Statistics tools

There are three main types of statistics tools available that can assist with providing a comprehensive overview of the usage that your repository is receiving:

Inbuilt repository reporting tools

Most repository platforms will have an inbuilt reporting and statistics tool. The tool is likely to record the number of visitors to each item, and the number of downloads of the files included in that item. It is sometimes possible to include the number of the item pages themselves so that users can see them. The statstics can often be viewed over time, for example to compare downloads over the past months or years. This type of raw, easy to understand, statistic is useful for advocacy as it provides easy to understand numbers that prove the effectiveness of the repository for access to the contained resources on the World Wide Web.

The repository will record item views by human users, and by search engine spiders that are crawling the web and indexing the pages. Unless the reporting can recognise and therefore not count hits from these spiders, the number of hits that the repository registers may be higher than is the case or real human readers who are making use of the repository contents.

The inbuilt repository platform statistics tools also usually report on internal statistics such as the number of users the repository has, how many item deposits have occurred, or what type the deposits are. These statistics are useful for reporting against internal targets for repository staff, and to show how much of an institution's output is currently being collected in the repository.

Web server log analysers

Each time a user visits a repository, a log is kept of their page accesses by the web server used by the repository platform. This information tends to be basic, but accurate. Tools are available that analyse these log files to create charts and graphs of the use of the repository. They log the IP address of the user, the date and time that they visited, what web page they viewed, and optional additional information such as the page they came from (what link they followed to the repository) and which web browser they used.

The statistical reports generated by these tools do not understand repository platforms, so cannot provide reports on the contents of the repository. They instead report purely on the visitors to, and their use of, the repository. As with the inbuilt repository tools, unless excluded by the system, the reports will include visits by both human users, and search engine spiders. The visits by the spiders are likely to skew the results and show an increase over the real number of human users.

Javascript analytics tools

The third type of statistic tool is provided by a centralised system which tracks visitors to your repository and reports the details back to the central system. When using such a system, you will be instructed to insert a few lines of Javascript code into each of your repository pages. This is typically accomplished by editing a common header or footer file that is automatically applied to all pages in your repository. When a user visits a repository web page, their browser will execute the Javascript which sends information about their visit to the central server. The central server can compile the statistics into statistical reports. Javascript analytical tools tend to offer very complex information about the browsing behaviour of the user, and can perform tasks such as showing their route through the repository from entry to exit. A typical route might be coming from a search engine, viewing an item, then browsing for further items.

The most well-know Javascript analytics tools is Google Analytics and some repository platforms are pre-configured to support Google Analytics. One big advantage of these systems is that they are generally not susceptible to registering extra hits from search engines, as a search enginge crawler is unlikely to run the Javascript that registers its view of the page.

Each of these methods has different strengths and weaknesses in terms of the statistics it can produce, and each provides a different view of the information. It is therefore likely that you will choose two or more of these options for your repository in order to be able to provide a comprehensive overview of the use of your repository.