Can you suggest a Log Analyzer?

Lois wrote:

 > "Chris Hope" wrote about Brucie's post:
 >: He's right and wrong. For most of those you can have some idea, but
 >: the numbers are never absolute, especially so on highly trafficked
 >: sites.
 >
 > I agree. 2 sets of stats tracking the same pages can produce different
 > results.
 >
 >
 > The
 >: most useful thing about webstats are looking at trend over time and
 >: where they came from (which in itself is not entirely accurate).
 >
 > They can give a pretty good idea. If you look at the examples of the
 > stats program I linked to, you'll see that you can click on a line for
 > each visitor and see where that person entered the site, how long s/he
 > spent on each page, and which page s/he went to next. Of course we
 > don't know if the person actually read the page before going to
 > another page, but we can get an idea of which pages are popular, where
 > visitors come from, and where they go.
 >
 > A better program for a site with a lot of traffic is phpOpenTracker.

All you can tell is when a file was requested.

Suggesting anything to the contrary will only perpetuate the myth that
you can determine user behaviour, particularly amongst noobies.


--
Charles Sweeney
<a style='text-decoration: underline;' href="http://CharlesSweeney.com" target="_blank">http://CharlesSweeney.com</a><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Can you suggest a Log Analyzer? 

On 15 Oct 2004 08:43:49 GMT Charles Sweeney <me.DeleteThis@charlessweeney.com>
broke off from drinking a cup of tea at No thank you to write:

 >All you can tell is when a file was requested.
 >
 >Suggesting anything to the contrary will only perpetuate the myth that
 >you can determine user behaviour, particularly amongst noobies.

But if you put a large enough price tag on the product, people will
believe anything.

For example. Let's say I offer a free of charge log analyser, people
assume "it's okay, but basic".

Now, if I offer the same product, but packaged in a shrink-wrapped box
for $5000 (sorry, but most readers are American) and claim it to offer

"professional web log file analysis"

people will assume "it's professional, it's great"

Like that man on the BBC the other week who couldn't sell cups of tap
water for 0.1 p per cup, but having bottled the same water could sell
it for far more!


Matt


--
If your encyclopaedia doesn't list "widget glass", you're reading the wrong encyclopaedia.
The Probert Encyclopaedia. Its not the same.
<a style='text-decoration: underline;' href="http://www.probertencyclopaedia.com" target="_blank">http://www.probertencyclopaedia.com</a><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Can you suggest a Log Analyzer? 

Lois wrote:

 > A better program for a site with a lot of traffic is phpOpenTracker.

Hmm interesting, especially given I'm a PHP programmer I'll check that
one out. Thanks.

--
Chris Hope - The Electric Toolbox - <a style='text-decoration: underline;' href="http://www.electrictoolbox.com/" target="_blank">http://www.electrictoolbox.com/</a><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Can you suggest a Log Analyzer? 

owen wrote:

 >
 > "Luis Veras" <veras.RemoveThis@tricom.net> wrote in message
 > news:b5GdnTs22cMNh_LcRVn-2w@codetel.net.do...
  >> Can you suggest a good and inexpensive raw log analyzer? I really like
  >> Webtrends, but I'm not prepared to pay $495 for a log analyzer, at
least
 > not
  >> right now. Any suggestions?
  >>
  >>
 >
 > I find it best not to analyse logs at all, just flush straight away
and
 > forget about it.

I only look at mine from time to time.

--
Charles Sweeney
<a style='text-decoration: underline;' href="http://CharlesSweeney.com" target="_blank">http://CharlesSweeney.com</a><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Can you suggest a Log Analyzer? 

"Lois" <auto-newsgroups.TakeThisOut@wordsweave.com> wrote in message
news:fIJbd.497$Ob.423@fe61.usenetserver.com...
 > "Chris Hope" wrote about Brucie's post:
 > : He's right and wrong. For most of those you can have some idea, but the
 > : numbers are never absolute, especially so on highly trafficked sites.
 >
 > I agree. 2 sets of stats tracking the same pages can produce different
 > results.
 >
 >
 > The
 > : most useful thing about webstats are looking at trend over time and
where
 > : they came from (which in itself is not entirely accurate).
 >
 > They can give a pretty good idea. If you look at the examples of the
stats
 > program I linked to, you'll see that you can click on a line for each
 > visitor and see where that person entered the site, how long s/he spent
on
 > each page, and which page s/he went to next. Of course we don't know if
the

Ok, I will bite - how can you tell how long someone spent on each page? AND
which page they went to next?

If you can answer those two questions accurately we can all make millions.<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Can you suggest a Log Analyzer? 

"Lois" wrote:
: > If you look at the examples of the stats
: > program I linked to, you'll see that you can click on a line for each
: > visitor and see where that person entered the site, how long s/he spent
: on
: > each page, and which page s/he went to next. Of course we don't know if
: the

"Bill Logan" responded:
: Ok, I will bite - how can you tell how long someone spent on each page?
AND
: which page they went to next?
:
: If you can answer those two questions accurately we can all make millions.

Here's an example from one of my stats pages:

---.vc.shawcable.net
.. First referrer: direct hit
2004-09-20 06:05:45 /portfolio.html
.. 2004-09-20 06:05:35 /home.html
.. 2004-09-17 05:05:19 /about.html
.. 2004-09-17 05:04:08 /publishing.html
.. 2004-09-17 05:03:20 /portfolio.html
.. 2004-09-17 05:00:24 /home.html
.. 2004-09-10 08:39:11 /home.html

The above stats are from before the site was live, so I know who the visitor
was; he was one of the few people who knew how to access the site at that
point. From the above, I can see that he went to the Home page on Sept. 10,
returned on Sept. 17 to view a few more pages, waited from 6 seconds to 3
minutes each time before going to another page, and came back on Sept. 20.

Those stats tell me how long his computer spent on each page. They do not
tell me whether or not he actually read the pages, or whether he was viewing
the window with my site open between clicks. They do show which site page he
went to next. I have a separate program to track clicks on external links;
there are lots of programs that do that.

I didn't think that was worth millions, but what do I know.

Lois
 >> Stay informed about: Can you suggest a Log Analyzer? 

 > What do you actually want to achieve?

Well, I'd like to see visual visitor trajectory and visual scenarios. I'd
like to see when visitors don't follow the site's intended trajectory, and
why. This way I can find out what's lacking in my content.

Of my referrals, I need to know who is pointing to which page. And I need
the referral list not to be limited to only 30 sites. I want to know if
those that are hotlinking my images are those that I actually authorized to
do so.

I want to see graphs with visitor volume trend throughout the days or weeks.
I want to see the actual pages (ie. html) volume, and not just the file
volume. I know that sometimes I have more requests, but because of the
hotlinks, and not because more people actually came to my site. I want to
see unique visitor count, and not just page requests.

And also anything else that might be good for analyzing my traffic.<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Can you suggest a Log Analyzer? 

On Sun, 17 Oct 2004 19:29:43 -0400 "Luis Veras" <veras.TakeThisOut@tricom.net>
broke off from drinking a cup of tea at to write:

  >> What do you actually want to achieve?
 >
 >Well, I'd like to see visual visitor trajectory and visual scenarios. I'd
 >like to see when visitors don't follow the site's intended trajectory, and
 >why. This way I can find out what's lacking in my content.

Interesting terminology. However, I think I understand what you mean.
It can not be reliably done. You could require password access to
every page, and track the path that way, but I think you might notice
a significant reduction in site usage <g>

 >
 >Of my referrals, I need to know who is pointing to which page. And I need
 >the referral list not to be limited to only 30 sites. I want to know if
 >those that are hotlinking my images are those that I actually authorized to
 >do so.

You can do this with Analog.

 >
 >I want to see graphs with visitor volume trend throughout the days or weeks.
 >I want to see the actual pages (ie. html) volume, and not just the file
 >volume. I know that sometimes I have more requests, but because of the
 >hotlinks, and not because more people actually came to my site. I want to
 >see unique visitor count, and not just page requests.

Not possible. All you can reveal from the log files are requests made
to your http server. Extensive "spy" systems may track some additional
data, but again not reliably.

Analog can report requests for any pages or combinations thereof made
to the server, from whcih you can see which pages are requested most
often from the server, which *may* with a large enough sample provide
*some* indication as to relative page popularity.

 >
 >And also anything else that might be good for analyzing my traffic.

The bank statement?

By which I mean, at the end of the day unless your server is
overloading, or your transfer allowance is being exceeded, all that
matters is the monies made.

http log files are provided for system administrators to monitor http
server usage and abuse, they are not provided for ego massaging.

Matt

--
If your encyclopaedia doesn't list "widget glass", you're reading the wrong encyclopaedia.
The Probert Encyclopaedia. Its not the same.
<a style='text-decoration: underline;' href="http://www.probertencyclopaedia.com" target="_blank">http://www.probertencyclopaedia.com</a><!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Can you suggest a Log Analyzer? 

You can try Web Log Storming:
http://www.datalandsoftware.com/weblog/

It will also allow you to explore your logs and to get into details
about sessions. Then you should decide can it give you valuable
informations about your website visitors.
 >> Stay informed about: Can you suggest a Log Analyzer? 

"Matt Probert" <comments DeleteThis @probertencyclopaedia.com> wrote in message
news:417369b4.6745018@news.ntlworld.com...
 > On Sun, 17 Oct 2004 19:29:43 -0400 "Luis Veras" <veras DeleteThis @tricom.net>

 > Interesting terminology. However, I think I understand what you mean.
 > It can not be reliably done. You could require password access to
 > every page, and track the path that way, but I think you might notice
 > a significant reduction in site usage <g>

I'm aware it can't be reliably done, but it does give you a pretty good idea
of the general trend. No matter how much real collection of what really
happens is lost, whatever information I can get is a good survey of what's
really happening.

  >>Of my referrals, I need to know who is pointing to which page. And I need
  >>the referral list not to be limited to only 30 sites. I want to know if
  >>those that are hotlinking my images are those that I actually authorized
  >>to
  >>do so.
 >
 > You can do this with Analog.

According to the demo online you can only see up to the top 30 referring
sites that link to your domain. But I didn't see any reports of any
referrers doing so to specific pages. Being honest, I haven't tried analog
yet, I just saw their online demo. So I may be missing some of its features.

 > Not possible. All you can reveal from the log files are requests made
 > to your http server. Extensive "spy" systems may track some additional
 > data, but again not reliably.

Again, if as a general rule the log can only record 70% of what really
happens, regardless, if it says I received 10,000 hits on Saturday and it
says I received 15,000 hits on Sunday, then it's obvious I get more traffic
on Sundays (even though it didn't count the actual 14,000 that came on Sat
or the actual 21,000 that came on Sun). See my point?

 > The bank statement?
 >
 > By which I mean, at the end of the day unless your server is
 > overloading, or your transfer allowance is being exceeded, all that
 > matters is the monies made.

Yes, that's true. But with analysis of visitor behaviour, I can find better
spots for my sponsor's links. I can see if my visitors like new additions to
the site. I can see if the new content I spent 2 weeks working on really
pays off, or if visitors are just ignoring it. With this I can see if the
new image I placed on this or that page, made more people actually click on
the next section or not. Etc, etc.<!-- ~MESSAGE_AFTER~ -->
 >> Stay informed about: Can you suggest a Log Analyzer?