PAN, Mambo, and AWStats - Statistics, The page counting differences explained. Options

[zimen]

This is the conclusions i reached about statistics of impressions.

I tried disabling beacon.
Tried java invocation and php invocation.
Tried the specific module.
Tried pasting code in template (java, php ecc...)

but still Awstats displays double impressions. Ok AwStats is not reliable. So i tried TFS and displayed 5000 in 24 hours (like awstats). Ok also that is not reliable.

So i did a test:

Unpublished the phpadsnew and published the banner from mambo's banner component, and guess what? I got same impressions as AwStats AND TFS !!! (PhpAdsNew does not go more than 2800)

So i think it's a bug with mambo and that unless I'm stupid and missed something out, PhpAdsNew stats are not reliable inside mambo.

Could someone shed some light over here please??

thank you

Simon

[Axe]

If you're calling the script via JS, then you will get double the hits.

Each mambo page load is going to load 2 pages...

Page 1 - The mambo page you're looking at
Page 2 - The PAN banner page that outputs the HTML for the banner (whether it be via JavaScript, iframe, or whatever).

If you've got 3 banners per page, then each Mambo page loaded is going to load 4 pages (mambo + 3 PAN script loads).

It's not a bug, it's correct. AWStats reports what's loaded in the log file.

Mambo's built-in banner system is hard-coded into the HTML output (just as PostNuke's is too, and many other scripts that have built-in banner functionality), and the HTML code to display the banner is directly in the page's HTML code (which you can see by viewing the source of the page).

So, there's no bug in PAN, Mambo, or AWStats, they're all reporting accurately.

I was running NKAds on one of my PostNuke sites, both scripts running on the same domain name. Even though the PostNuke site was reporting 30-40K unique page views per day (2 banners per page), AWStats was reporting 100K or more page views per day (because each of those JavaScript calls to NKAds was another page load as far as the log file is concerned - which is correct). Same thing would happen with PAN.

This is why I now have a single copy PAN running on a completely separate domain name from all my other websites. All of my sites now call PAN on this separate domain via JavaScript, so that AWStats is not including them for their respective site logs.

The separate domain doesn't actually receive more than a couple of hundred human visitors per month (I'm surprised it gets that many) for content actually on that domain, but over the past week or so, it's served over 100,000 pages every day, to several thousand site visitors. These aren't people going to that site to look at content, it's merely URLs being called to display ads. It's accurate.

[zimen]

Thank you Axe, you havee been very clear and i understood perfectly. But i still have one doubt and would be grateful if you could answer:

The other day, i totally disabled phpads. I put a banner via the default mambo banner component, and checked the impressions it made over 24 hours.

After 24 hours it gave me 4500 impressions. PAN the day before gave me 2800, and that was the maximum PAN ever reached, whilst normally from mambo banners i get between 3500 to 5500.

In this case, having disabled PAN, i should have only 1 page load for every mambo page load, not 2 (mambo + pan).

So why is that? Am i missing out something else?


Thank you very much, please do answer me as i sold already and adv page to a big client.


Simon

[Axe]

Unless Mambo is also using JavaScript to call its own banners module, I'm not sure.

I've very little experience with using banners in Mambo, as I've always used separate banner scripts anyway (same with PostNuke, although I know for a fact PostNuke's built-in banner system outputs directly to the HTML).

That's part of why I always run separate banner scripts though, it helps weed out confusion, and a dedicated application like phpAdsNew is going to give much more accurate information, and much more reliability when it comes to probability of banner display, than any CMS's built-in banner systen module. Not many of them (if any) have had the same amount of work put into them that PAN obviously has.

[zimen]

Axe,

I understand. I will use pan to see banner stats.

I did a new test in the last 12 hours:

1: Installed pan on a different domain than the website it's serving.
2: I placed the pan java invocation code INSIDE the mambo banner component, so i could view BOTH in the same time.

The results are that after 12 hours mambo shows 2175 impressions and pan shows 1098. Double!

Mambo banner does not use java but i have the wysiwyg editor wich does. Could it be that?
view the source code of the website: www.pigmag.com

also, is there another stats software around wich uses beacons or similar, wich i can download?


thank you again!

Simon

[Axe]

Yeah, it could also be something else that's being called. Again, not entirely sure about Mambo, but with PostNuke (at least, most versions of it), two .js files are called with each page that's loaded.

CODE

<script type="text/javascript" src="javascript/showimages.php"></script>
<script type="text/javascript" src="javascript/openwindow.php?hlpfile="></script>

So, for each PostNuke page loaded, that's 3 page views as far as Apache's log file (and thus AWStats) is concerned.

I would imagine it's something similar to this (possibly the WYSIWYG editor, as you suggest) that you're seeing with Mambo.

[zimen]

Axe,

Thank you very much !!

Finally I understood. And this makes me happy!

Well, on one side less because i thought i actually had that number of page impressions instead it turns out i have much less! but never mind, at least i'm happy i finally understood

i think this should be written somewhere in a f.a.q or in a sticky topic because i didn't find any posts that mention javascripts counting as impressions, and there are alot of posts about the problem, over here AND on the mambo forums.

From what I understood, if I have access to the awstats i can configure it to "not" count those doubles right?


Simon

[motowebmaster]

Just to add my comments, as an experienced PostNuke user running PAN on a separate Domain/IP.

For CMS Portals, you have to allow some latitude on the information an Apache Log would collect vs. what actually is considered a page view on your site. As previously suggested, use Httpd Log hits and page views just as mechanism to gauge the load on your server; it will never be an accurate count of actual user page views.

I use a PHP Function to update a MySQL database with basic info, based on the original PostNuke Module. It collects basic info, but it's useful for my purposes.

Hope this helps,
Shawn

[ccie9277]

Just to be clear on this. I run this in joomla and it counts the same if I hard code the java into the page or if I run it inside of the banner software. I've noticed that even google analytics will show twice the impressions that pan does.

[FixedGearFever]

Okay, I understand why awstats is so much different from my phpnuke page views. That's not a problem. I feel fairly certain that my pageviews are quite accurate (in phpnuke) as I have disected the code and checked how the table is updated....

Why is it that PAN is not yielding a similar number for a banner that is on every page?

-jsp