Wrong statistics?

[Gryzor]

Hello everyone,

Today I did my annual stats analysis and I saw something weird: I'm running a forum and a mediawiki installation as parts of our site, and Google Analytics gives me 1.3m pageviews for 2011. However, the SMF stats show a staggering 16.300.522 pageviews for the same period! It's important to note that the GA figure is for both the forum and the mediawiki (which is quite popular, too).

Last year there was a similar problem, with GA reporting a total of 917k pageviews, and SMF alone reporting 1.68m.

Am I missing something here or is there something wrong?

Cheers
T
www.cpcwiki.eu

[Gryzor]

[bump]Anyone?[/bump]

I'd expect this would raise an eyebrow or two...

[Illori]

i believe that google analytics does not always count all the bots/search engines that browse your forum while smf does.

[Gryzor]

This could be a good explanation, but... it would also mean I had 15m pages served to bots?? I'm not sure, but it kinda sounds too high...

[vkot]

I have the same problem: Pageviews count in SMF statistics are more than 10x the pageviews count in Google Analytics! It's a forum with a long history, 15 years, and high in search engine ranking.

Can it be that more than 9/10 of my pageviews are from search engine bots? I find it outrageously high. How can I analyse this?
Is it common in forums to have such a big difference between SMF statistics and Google statistics?

[margarett]

Actually I think that it's just the algorithms which are different.

SMF counts every single page view (except some known stuff) as a hit. GA probably (just an idea, not sure that this is true) "filters" pageviews by session or IP and for sure doesn't count its own page views.

Code Select Expand

// Do some logging, unless this is an attachment, avatar, toggle of editor buttons, theme option, XML feed etc.
if (empty($_REQUEST['action']) || !in_array($_REQUEST['action'], array('dlattach', 'findmember', 'jseditor', 'jsoption', 'requestmembers', 'smstats', '.xml', 'xmlhttp', 'verificationcode', 'viewquery', 'viewsmfile')))
{
// Log this user as online.
writeLog();

// Track forum statistics and hits...?
if (!empty($modSettings['hitStats']))
trackStats(array('hits' => '+'));
}

[a10]

Don't know about how Google calculates. On my forum I blocked some overactive bots (bing, some useless leeches like ahrefs etc, using combination of htaccess + robots.txt), smf daily page views immediately went down by approx 50%.

[vkot]

But... if hits were filtered by session or IP, they would be called "unique visitors", not "pageviews". Even if it was like this, in my forum the average pageviews per session is low (2+), in no way it's 10+, it doesn't make sense.

I don't want to block any search engines, unless I had serious server load problem, which I don't.

[margarett]

As I said, I am just guessing, no idea how GA works (I do know how SMF works, it's in the code )

[vkot]

I was embarassed because I was advertising that my forum has X million pageviews per month but when I installed Google Analytics it showed X/10. Now, I don't know what to say to people who ask me...

[Suki]

Server and client side metrics will always differ from each other. Client side metrics will most likely always rely on JS, this means any bot will simply not be taken into account as the vast majority of them doesn't bother with interpreting JS, some of them doesn't even bother to interpret HTML too.

Theres also things like ghostery which actively refuses GA code, users who are using it will not be counted by GA.
Users with JS turned off will not be counted as well.
Old mobile browsers will not be counted either, yep, those still exists, I was using one as recently as early this year.

Theres is also the way GA do its own metrics and what they believe to be a "valid" visit or pageview, SMF doesn't really care or put that much effort into it, as long as anything gets executed until the very moment writeLog() its called then it gets counted as a pageview.  GA will surely be a lot more picky when it comes down to tracking views and visits.