log analysis: Myths, Truths and Tips
By John Marshall
Since servers started collecting data, web marketers
have been right there on their heels. Much like gathering demographic
reports from television and print campaigns, marketers wait
next to the printer, getting the latest data from their website
hot off the press. Standing, looking over the shoulder of the
tech guy, web marketers attempt to glean information from a
static stream of numbers and charts.
There are many myths about log file analysis. But by tweaking
both the software and the mindset of the software's user, getting
the most out of log files can become a more efficient, effective
process. The following myths and tips highlight the most common
misconceptions and offer suggestions on making things work better:
Web log analysis myths and tips:
Myth: It takes a long time to analyze website
data. Log files are so large and the software needs to go back
and forth to the server to gather all the information it needs.
Tip: The truth is, many web log analysis packages
are slowed down by something called Reverse DNS lookups. The
Reverse DNS function is what generates the information on visitors'
geographic location, e.g. whether a person's coming from a .fr
(France), .jp (Japan) or .de (Germany) address. Unfortunately,
that geographic information is typically so unreliable that
it's not worth having. Switch off the Reverse DNS function in
your log analysis software to speed up the process.
Myth: The more reports, the better. Web log
analysis software typically offers a slew of reports. Everything
from 404 errors to browser type, screen resolution and top-level
referrers are available, along with more detailed information
like length of time to a particular page, numbers of users who
exited from a particular page, and comparisons of campaigns.
Tip: Not every report is of interest in every
case. Instead of looking at the mass of reports every time,
utilize the filters that are typically built into the software
to segregate the data in which you have an interest. By filtering
results, web marketers avoid becoming overwhelmed by the data
that doesn't matter, and are instead able to focus on the task
at hand--understanding how visitors interact with their web
Myth: Web log analysis software needs to work
with an external database to be effective, and because of the
tedious nature of databases, the software shouldn't be installed
by anyone without an IT certification.
Tip: External databases seem innocuous, but
tend to be overkill for many websites. Additionally, external
databases are more complex to implement and maintain, driving
up both time and monetary requirements. Unless you're dealing
with a hugely complex site, avoid the database-and by virtue,
avoid calling in the IT cavalry.
Myth: Web marketers should be highly interested
in what visitors are doing in real time. Getting an immediate
view of web site activity is beneficial.
Tip: Don't succumb to voyeurism. Real time
information is fascinating, but distracting. Web marketers'
time will be better spent drawing conclusions at the end of
a set period of time, whether that is half a day, a day, a week
or a month. Which would give web marketers a better idea of
the life of an elephant living in the zoo? Watching the elephant
move around for five minutes or reviewing a summary of his actions
for the entire day? Look at the data on a regular schedule,
but don't get make the mistake of confusing 'interesting' and
'useful'. Beyond it being a waste of time, these gimmicks and
gewgaws also consume CPU bandwidth.
As in many other aspects of life, being able to separate the
myths from the truths is important to success. By following
the tips mentioned above, web marketers will be better able
to understand visitor interaction on their website.
Former Netscaper John Marshall is CEO of ClickTracks, developers
of innovative web metrics software ClickTracks Analyzer. John
can be contacted at firstname.lastname@example.org
or by visiting www.clicktracks.com.
ClickTracks Analyzer recently beat both WebTrends and SiteCatalyst
in ClickZ's 2003 Marketing Excellence Awards in the 'Best Web
Site Analytics' category.