Freelance at Emarketing Manager
24 June 2005 09:15am
I wanted to ask users about web log analytics. I have had to use two different logs between an old and a new website. The differences in hits/impressions and visitors tell a vastly different stories. I have read elsewhere that using two different logs can display different pictures of web traffic. I am presently using web trends and web funnel analyser. I wondered if users had a preferably free good web log analyser and some hints on how to deal with different readings from their experience.
Thanks for your help in advance.
Producer at eMetrics Marketing Optimization Summit
27 June 2005 01:21am
On 09:15:47 24 June 2005 helruna wrote:
some hints on how to deal with different readings
from their experience.
"Don't try" is my advice.
Web analytics tools capture data in multiple ways. They cleanse that data in multiple ways. They report out that data in multiple ways. While they are consistent, they are never in agreement. The business side of the house demands precise numbers. You'll have to show them how tracking trends is more important than knowing specifics to three decimal points.
Director at ISSEL
27 June 2005 10:29am
Before you spend lots of time worrying about trying to get the figures to agree step back a pace.
Ask yourself the question – why do I need the figures and what am I trying to find out?
So long as the figures from your tool are calculated consistently that’s a first step because that will allow you to establish trends. You need to identify your objectives for the site in the context of the business, creating a clear view of what you are trying to achieve and setting targets. This will help you talk to to your less online focused colleagues to explain your contribution to the business as a whole. It may even help you underpin that business case for ongoing investment.
If you are concerned about accuracy then why not use the ABCE definition of traffic as guidance. You should also look to be able to analyse the data that ABCE doesn’t allow separately as that can be very useful for technical/error analysis and how well spiders are able to index your site.
Colin Cooper email@example.com
ISSEL – Pilot Software
CEO at Econsultancy
27 June 2005 13:39pm
Is that 2 different sets of log files or 2 different web log analysis tools? Or both?
The log files, if generated by the same web server using the same configuration, should generate consistent log files. I say 'consistent' because they may have a lot of redundant information in them which needs stripping out of any reporting, but at least you can compare like with like.
If they are not the same web server, or the settings have changed, then this could be the root of your problem. For example, have you moved from CLF (common log file format) to ECLF (Extended common log file format)? Have you kept your session time out setting the same on the web server (usually 30 mins)? These things, and others, could make a big difference to what results will get spit out of standard like-for-like analysis and reporting.
If you're sure that the logs files are consistent, then the most likely cause of difference in figures between 2 analysis tools is the way they are configured. Again, check the session timeout assumed. Most likely, though, it will be an inconsistency in what is being filtered out or not. Typically, around 15% of site traffic might be non-human and needs to be filtererd out. If you've got frames, or automated agents pinging your site every 10seconds, or a range of other things, your stats could be skewed a lot if your exclusions are not thorough and consistent.
As Jim says, you should focus on trends and shouldn't worry about small inconsistencies. However, very large inconsistencies are most likely down to one or other of the reasons I mention above.
We've got multiple (3) web analytics tools running on our site, using page tags rather than web server log file analysis, and the margins of difference between them is actually so slight as to be not worth worrying about.
27 June 2005 13:49pm
Thanks for the advice guys!
28 June 2005 12:38pm
p.s. in answer to your question about free tools out there, perhaps the most common are:
Statcounter - for free page tag / ASP toolAnalog - for free log file analysis
There are downsides to these free tools and with prices coming right down on professional web analytics solutions (e.g. to around £100 a month) then it might well be worth looking at these too.
Plenty of info on these tools in our Web Analytics Buyers Guide 2005
Director at Velo//
30 June 2005 09:28am
And in addition to Ashley's two, I've been using net tracker lite. It is available for free download from here:
I've been using this software since the start of the year, and had no problems
Hope this helps
... a unique online digital marketing agency that specialise in inbound and outbound email, fax and SMS broadcast solutions
Free market research on digital marketing
Daily Pulse: award winning newsletter
It takes 30 seconds to register