Mach5 Support

FastStats Filters Tutorial

Filters are a powerful tool for analyzing your web server log files. The best way to learn about filtering is to dive right into it. This tutorial will give you example uses of filters that will help you enhanced your web site.

But before we talk about all of the novel uses of filters, you should know how to add, edit, and remove filters from a report. In the Report List Dialog, click Edit Report. Click the Filters tab. Click Add and then begin reading this tutorial.

A filter can either include or exclude data. An include filter is good for analyzing only a certain subset of data (for example, only Internet Explorer 4.x users). An exclude filter, conversely, is useful for excluding a subset of data from the analysis (for example, excluding yourself from the site statistics). If you specify multiple include filters, the data will be included if one of those include filters applies. Similarly, if you specify multiple exclude filters, the data will be excluded if one of those exclude filters applies. You can mix both include and exclude filters.

The following is a list of all filter types and examples of how to use them in your own reports.

Filter by Domain Name

1. Exclude ‘local’ hits. Use wildcards to specify all IP addresses that may have been used by internal users in testing the web site. This allows you to measure only the visitors to a web site, not internal web site users. For example, to exclude every user from MIT, specify the following wild card: ’18.*’. If you do not know your local IP addresses, ask your system administrator or ISP.

2. Measure Intranet usage. This essentially the opposite of the ‘Exclude local hits’ filter. This time, you only want to include local hits. This lets you measure the usage patterns of internal users.

3. Track the users from a specific domain. Find out if your competitors are accessing your web site! Just set FastStats to only include requests from your competitor’s IP addresses.

Filter by Requested File

1. Zero in on the performance of a specific file or directory. Specify a filter for a file or directory and include or exclude that data from your report. For example, some areas of your site may “drown out” less popular areas. Include only the less popular areas and analyze them (or exclude the more popular areas and then analyze those).

2. Exclude toolbar buttons from the analysis. Certain files, including toolbar buttons, may be requested disproportionately more than other files and may, again, “drown out” the other files. You can exclude these files from the analysis.

Filter by Range of dates
Narrow down the information included in your report.

Filter by Day of week
Do your hits always drop off on weekends? Find out why by only including hits from Saturday and Sunday.

Filter by Hour of day
Your web site may attract a different crowd at various times of the day. For example, web site with weather reports will probably receive an increased amount of requests in the morning, when people are checking the forecast for the day, and in the afternoon, when people are checking tomorrow’s forecast.

Filter by Referring URL

1. Track an advertising campaign. Include only referring URLs from the site you’re advertising on. This will track which pages on your server (and which pages on their server) users are responding to the advertisement.

2. Track all links from a specific web site. Do you have a partnership with another web site? Just want to see which pages on a specific web site link to your web site? Filter the referring URL by any web site.

Filter by User’s Operating System
Different operating systems may mean different audiences. Macintosh, Linux, and Windows may be interested in different content on your web site.

Filter by User’s Browser

1. Browser specific information. Are Netscape or Internet Explorer users avoiding a particular part of your site because it does not work well with their browsers? Or, if you have a large web site, conduct your own browser market-share test.

2. Text only information. Include only requests from Lynx or other browsers.

Note: many browsers, such as Internet Explorer and Opera, report themselves as Mozilla (Netscape) compatible; it is difficult to include only Netscape users or exclude Internet Explorer users. Search the FastStats help for more information on this topic.

Filter by Web Spider
Find out which pages web spiders (the automated computer program used by search engines to index your web site) have reached (and which pages they have not reached).

Tips and Tricks

Having trouble using FastStats's filtering features? This page has some tips and techniques that should help you become a filter power user in no time.

A Short Explanation of Filters
Filters are a way of restricting the data that is included in your report. Want to only view statistics for the month of October? Use a filter. Want to analyze the patterns of only Internet Explorers as they move through your web site? Use a filter. Similarly, filters let you track the hits coming from one referring page or domain (useful for tracking an advertisement), or track people coming from the AOL.COM domain.

Include or Exclude
Include and exclude filters do just what the say. Let's say you set up an include filter that includes all Requested files matching /dir/*. This will do exactly what you expect -- only requests for filenames that match http://www.website.com/dir/ will be included in the FastStats report. Similarly, if you set up an exclude filter on the /dur/* directory, any file that was not in the /fast/ directory will not be included in the report. What if you set up one filter that includes /dir/*, and one filter that excludes /dir/file.html. Any file that is in the /dir/ directory will be included, except for /dir/file.html.

What If I Have More Than One Include or Exclude?
This is where it can get tricky. Let's say we have two include filters set up, one for /dir/* and the other for /otherdir/*. What happens? Well, it depends on the setting of some FastStats options. Click the Report menu and choose Option. On the first tab (Log File Options), see whether Only include log file records if all include filters are true is checked. If it is, then a file has to match both /dir/* and /otherdir/* to be analyzed. It this option is not checked (the default setting), then what you expected will happen -- files in both /dir/ and /otherdir/ will be included in the analysis. The same thing applies to exclude filters and the Only exclude log file records if all exclude filters are true option.