If you want Web stats for an departmental site please send e-mail to:

web-consult@virginia.edu

Frequently asked questions about webstats

How long before the stats start working?

The stats start being collected in the middle of the night the first day you sign up. Stats will keep working until you stop them. It's all automatic.

Where are the stats?

The Web Stats software creates a directory in your public_html folder called stats. If your userid is something like mst3k, the UNIX path to your stats files is:

/home/mst3k/public_html/stats/

If your userid is mst3k, the URL of your stats is:

http://www.people.virginia.edu/~mst3k/stats/

(note the ~ tilde character before your userid - this is always there for your URL on www.people)

There are up to one month's log files saved in this directory, along with the monthly analysis files. The analysis files (Jan.html, Feb.html, etc.) are kept for 12 months until they are overwritten. The log files are only kept for the current month.

Can I get stats on both my departmental pages, and sub-department pages?

Yes, although stats cover an entire directory and all the subdirectories of the Web site. You will have to e-mail web-consult@virginia.edu to get signed up. Be sure the e-mail includes the URL of the directory or directories that you want to track. Just browse to the directory, and copy the URL, then paste the whole URL into the e-mail.

Where will my department or sub-department stats be located?

Departmental stats are always in the directory being tracked, in a new directory called

dir_stats

If your department were called "Modern Literature", and your site was on the main U.Va. server (minerva.acc), and your account was modlit, your stats would be:

http://www.virginia.edu/~modlit/dir_stats/

Or perhaps you have a newer Web site without the tilde (~) character:

http://www.virginia.edu/modernliterature/dir_stats/

Can I get stats for my site on faculty.virginia.edu?

Yes. The sign up process is the same as for departments. E-mail web-consult@virginia.edu with your complete URL. The stats will also be in a dir_stats directory. If you project was called "Benjamin Lab", your Web site stats would be in:

http://faculty.virginia.edu/benjaminlab/dir_stats/

Webstats reports

Why isn't more individual detail in the logs?

It would be nice to say that customer number 300 visited 10 pages within your site, and visited them in a certain order. This isn't possible without the use of cookies and special Web stats software. Cookies are often considered an invasion of privacy because they allow tracking of customer's behavior. Without a technology for tracking (like cookies) the World Wide Web is what is called "stateless". There is very little information about customer's current "state". You cannot know where they have been, or how many minutes they have been there. In fact, you can't even verify that they finished loading a page.

Typical sections of an analysis of the monthly state (for example Jan.html)

Your number of 'hits' is the third line in the first section at the top of the analysis.

Total successful requests for pages: 1234

Reports

Monthly Report: Number of pages loaded. We only analyze one month at a time, so this doesn't tell you much. There isn't disk space to save the log files for longer than a month.

Weekly Report: Number of page loaded by week. Also not very informative.

Domain Report: What kind of domains your customers visit from. Mildly useful. You can tell whether people come to your site mostly from .com (commercial) domains, or from U.Va. and other schools (.edu domains), as well as roughly how many overseas visitors you have.

Request Report: Which of your pages was requested, ranked by number of hits. This is really useful. This is a popularity list of your Web pages.

Referrer Report: Where your pages are linked from. This is very useful. Most of the hits will be from within your site. Scroll down the list and you'll see references from the search engines. Something like:

http://search.yahoo.com/bin/search?p=volkswagons+repair

This means that someone searched at Yahoo for "volkswagons repair", and found a link to your site. The + sign is HTML cgi for a space (since spaces are not allowed in URLs).

Error Report: Not very useful.

Browser logs

These logs are extremely large, and are only kept for the current week, and only for the entire host, not for individual sites. The output is very messy since Internet Explorer kind of calls itself Mozilla, when Mozilla is really Netscape Navigator. Each browser also identifies itself with details, including all the minor versions. The only use of this data is to see how many people are using new versus old browsers. Note that a number like 200 hits coming from Netscape Navigator 2.0 for Win95 is probably just one person hitting the server 200 times during the day.

The user.cfg file

You can edit the user.cfg file in your stats directory. Be careful, and make a backup copy before starting. You'll have to read the Analog documentation to know what is possible to do, and how to do it.

http://www.analog.cx/

Why can't I track individual customers and why aren't the stats more accurate?

The best that any Web site usage software can provide is a ballpark figure. This has to do with the nature of the Internet and the WWW. As Stephen Turner says, anyone who claims features like individual user tracking is probably not telling you the whole story. You can get the fascinating details at one of Stephen's many pages (this link is to a mirror site in Pennsylvania):
http://www.web42.com/analog/docs/webworks.html

The actual analysis is performed by the Analog program, which has documentation at http://www.analog.cx/

You can attend one of the Web Promotion seminars. The schedule is at:
http://www.itc.virginia.edu/training/webcert

Many thanks to the creator of Analog: Stephen R. E. Turner
Email: sret1 at ntlworld.com