AustLit records the following basic access counts by subscribing user by year and month:

• author searches
• work searches
• searches with nil results (i.e. found nothing)
• returned results (i.e. count of author or work summaries shown as search results)
• author detail views
• work detail views
• total page views of pages requiring subscriber identification (i.e. non public pages)

Subscription administrators may specify year and month ranges, and output format of usage reports.

The default format is HTML, but comma delimited text output is also available for customers who wish to use the information for detailed statistical analysis, graph production, etc.

Access to the statistics page is only available to a subscription administrator userid with the 'Accounting' access privilege, which requires a userid and password. Please contact

info-austlit@austlit.edu.au

AustLit can identify users by either:

• an explicitly provided userid and password, or
• the IP-address used to connect to the AustLit web server.

AustLit subscribers nominate how they want their user(s) to be recognised, and can use both methods for the same or different userids. For example:

• user FRED1 will be automatically bound to a connection originating from an IP address starting with '10.83.3.'
• user FRED2 will be bound to a connection only if the userid/password 'FRED2/quiet' is provided
• user FRED3 will be automatically bound to a connection originating from an IP address starting with '10.83.4.', or if a userid/password of 'FRED3/secret' is provided from any IP address

Subscribers may divide their users into groups to help break down and group statistical summaries.

Although automatically binding users to IP address ranges is simple, it does cause some problems:

• if 'off-site access' is required via public ISP or a non-subscriber owned IP address range, a password will be required to identify the user
• subscribers typically route all web traffic via web caches/proxies, which hide the end-user's real IP address. The request to the AustLit web server appears to originate from the subscriber's web proxy, and hence users in different parts of the subscriber's IP address ranges cannot be differentiated by IP address, requiring the use of a password.
• As userid/passwords are difficult to manage and inconvenient to users, multiple groups and users per subscriber (although fully supported by AustLit) should only be used when necessary.

As an alternative to a subscriber defining multiple AustLit userids in these cases, reports on the subscriber's own proxy logs may reveal sufficient statistics to determine usage by different groups within the subscriber's network.

Interpreters of the statistics provided should be aware that the analysis of statistics is often subject to misinterpretation, and that the analysis of web statistics is especially problematic due to the role of browser and organisational caches and the stateless nature of web 'sessions'. While AustLit's counts of searches, results and views are sound, we suggest that the count of 'total page views of pages requiring subscriber identification' be used with caution, and perhaps only as a 'trend figure'. For a discussion of these issues see:

• How the web works
• Why web usage statistics are (worse than) meaningless.