Unless you are a system administrator by profession, have the right tools or know your way around some text-processing UNIX tools (awk, sort, grep in particular ) , analyzing web server log files can be a daunting task.

Without command line kung fu, log files are not fun to look at.

Tantillus

Tantillus is web log analytics application that parses and analyzes Apache server log files to retrieve metrics that are beyond the capability of regular javascript based applications (page tagging) like Google Analytics, which have an inherent limitation: their limited scope and reliance on javascript. With page tagging some things will pass by unnoticed, including, but not limited to:

• Search-engines
• Visitors who have disabled javascript (albeit relatively few)
• Failed requests (client and server errors, such as 404 pages)
• All content which you can’t embed javascript into.
  (That’s everything but HTML files)

Scroll down to the end of the page for a DEMO link.

Features

• ONE-STEP setup: you only need to know where your log files are stored.
• Statistics in real-time. No periodic snapshots. No database.
• Perform tedious tasks that would otherwise require SHELL access (and adequate UNIX knowledge)
• A fast module that uses SQL-like query-language to filter and extract data.
• Play back accessed files, user flow and traffic sources hour by hour.
• Monitor your traffic LIVE: watch as traffic flows into your site.
• Easy to tweak and configure (uses an XML file to store config data)
• Unveiling previously hard to acquire information.
• Filter and visualize visitor activity and patterns of use.
• Very fast. Runs faster than Speedy Gonzáles, at least hypothetically.
• Easy configuration (uses an XML file to store config data)
• Content segmentation: group your web files into related segments (like images, scripts, videos, etc.) and get a precise measurement of much bandwidth (size) and hits (number of requests) each of these segments received as a whole (and the share of each file type)

SQL-like query syntax

Tantillus comes with an ad-hoc analysis and filtering tool which uses a querying language that’s very much close to, but not identical to nor as refined as, SQL. Supports filtering, ordering, grouping, aggregating, regular expressions, among other features that you can get your stuff done with.

Since we are not importing data anywhere (no database, remember?), but rather monitoring log files as we go, analyzing multiple websites is a breeze, you can run a query like this sample:
SELECT file,referer FROM /home/myname/logs/mysite.com/access.log WHERE file LIKE %blog% AND referer STARTS WITH http://www.google ORDER BY file without having to modify anything, except the log path. Go ahead and test-drive this in the demo. Elaborate usage examples and HOWTO’s included.

Find the needle in the haystack

Content filter: If you don’t speak SQL or don’t feel the need to, there’s a point-and-click content filter you can use to dig deeper into your site.

Know thy audience

Easily determine bandwidth and pageview distribution (depth of visit) among your visitors , hour by hour. Some visitors will, inevitably, leave immediately upon arriving at your site (bounce rate), pinpoint where this type of traffic is coming from, and which search keywords and referring sites caused it. Most users often leave (bounce) if they don’t find what they are looking for, or realize that the found site doesn’t conform to their expectation.^[note]

Multi-dimensional user data analysis:

Figure: each bubble represents a unique visitor (ip address). Notice how few visitors with a small footprint, pageview-wise, have a relatively large bandwidth consumption (the big yellow/orange bubbles at the right bottom)

Devide and conquer:
Easily track documents and content types .

Figure: The pageview/bandwidth ratio can be applied to directories or files as well. Each segment contains a pre-defined set of file types.

Gallery of data visualizations

Above: Compare this graph from a video-heavy site with..

..this graph from another site with no video content. Notice how the content distribution varies remarkably.

DEMO

A work in progress.. Release date to be announced