I have noticed that if you're not careful with your tags in
your templates then you can easily end up with the same page being
pulled for various URLS. If you've been lazy with your template design,
then it's impossible jsut looking at the URL to work out whether you're
on a section, article or issue page (you can't assume that just because
the template is called 'section'...). Either you have to add some
intelligence to the tracker so that it can ignore irrelevent URL
parameters, or you have to be careful about your template programming. I
think the latter is better.

As for multi-page articles, this is an instance of a more general
problem which you're going to want to tackle for the tracker.

Take the case of issues and articles. A number of alternatives are possible
* person visits the issue home page
* person visits a various different pages withn a specific issue
* person visits a specific article in a specific issue

Questions you are going to want to answer:
* how many people read (any part of) this issue?
* how many people read the home page for this issue?
* where did people click through to after viewing the home page for the
issue?
* which articles got the most page impressions in this issue?
* which issues got the most page impressions?
* number of articles seen by distinct viewers, issue by issue

I think that if you can do all of the above then it will generalise to
similar questions on the article and page-within-an-article level, ie
* how many different people read (any part of) this article
* how many different people read a specific section of this article
etc

BTW, http://www.offermatica.com/?campaign=gabt -- A friend recently
pointed me at this site, which points the way to some pretty nice, more
advanced use of website stats. Basically it lets you do little "change
and measure" experiments on your website. You set your website software
to randomly serve up one of a number of variants of a given page to each
of your users, then over time you can collect statistics on the varying
behaviour of people viewing the differing pages. This lets you be a bit
scientific with the results of the website stats, since you can test
every change before deciding to stick with it. Example: maybe you are
trying to increase the number of subscribers. Where's the best place to
put your promotional message, and which promotional message gets the
most clickthroughs? Normally you would just have to guess, but with this
kind of approach you can test the alternatives and settle on the best one.

Obviously it's not a trivial piece of software, however I thought I'd
let you know about it because then you can keep an eye open to allowing
this stuff to be implemented down the track.

How is the tracker being integrated with Campsite? Is it via server logs
or php code, or via a web 'bug' or javascript etc? phpopentracker allows
you to use either a little PHP fragment or a web bug as I recall.

Cheers
JP

Aleksandar Brajanoski wrote:

>sebastian came up with something i didn't seriosly think of before.
>so here is the converstation we had yesterday (i'll exclude the porn
>parts). are there industry standards here? what do you guys think?
>thanks.
> seb45t14n (3:22:01 PM): wanna clarify some tracker behavior
> mediaturtle (3:22:10 PM): yes mediaturtle (3:22:16 PM): go on
> seb45t14n (3:22:43 PM): the current one is: if the same url is
> requested twice in one visit, it will just count as one
> page-impression seb45t14n (3:23:09 PM): but, if same article is
> displayed using another url, it is counted twice mediaturtle (3:23:51
> PM): ok... seb45t14n (3:25:00 PM): for e.g. if section.tpl is
> requested with or without NrArticle, it will always display section,
> article doesn't matter for campsite seb45t14n (3:25:00 PM): but for
> the tracker. seb45t14n (3:25:31 PM): thatswhy in tracker_section.tpl
> NrArticle is always 0 mediaturtle (3:25:38 PM): i think i can fine
> tune this thru templates later on, if i decide to count visits to the
> same article thru different urls as one page impression. right?
> mediaturtle (3:25:47 PM): yes, yes, i understand mediaturtle (3:25:52
> PM): that is how i want it seb45t14n (3:26:29 PM): i have seen you
> have lot of articles with more than one page mediaturtle (3:26:54 PM):
> yes seb45t14n (3:27:07 PM): would it be best to track those just as
> one page_request, if user reads all 3 pages? mediaturtle (3:27:16 PM):
> the thing is article impressions vs. page impressions mediaturtle
> (3:27:37 PM): that's what's troubleing you? seb45t14n (3:29:12 PM):
> lets say other way: would it be right to track just diff. parameters
> as diff page_impressions? seb45t14n (3:30:12 PM): if your marketing
> guys are interested which article/section/topic is read a lot or not,
> this would give the information. if they like to have big numbers of
> page-impressions, it would not mediaturtle (3:34:56 PM): hmmmmmmmmmm
> mediaturtle (3:35:00 PM): let me think mediaturtle (3:38:09 PM): in
> ideal world you would have both things counted: how many times an
> article was read, but also you would count all page impressions no
> matter how many pages that article has. seb45t14n (3:39:58 PM): in an
> ideal world.... mediaturtle (3:39:59 PM): in other words, if i wanted
> to see the list of most read articles, i would have a tracker section
> where it doesn't count the page impressions, but one visit to an
> ARTICLE would create one ARTICLE impression no matter how many page
> impressions it produced. on the other hand, all page impressions would
> be counted in total page impressions for the site. something like
> that... mediaturtle (3:40:03 PM): yes... mediaturtle (3:41:22 PM): i
> think we should make it as it is now --> all page impressions count
> even if some of them are result of one visitor reading multiple pages
> of the same article
>

------------------------------------------
Posted to Phorum via PhorumMail