GoFuckYourself.com - Adult Webmaster Forum - Google Analytics vs Server Stats

- Fucking Around & Business Discussion (https://gfy.com/forumdisplay.php?f=26)

- - Google Analytics vs Server Stats (https://gfy.com/showthread.php?t=1077405)

Google Analytics vs Server Stats

there's a big discrepancy between GA's stats and my own server stats - GA shows 30% less visitors than server stats.

normal?

Of course it's normal. Server stats show you the exact amount of hits while GA shows only those that have JavaScript enabled (supported).

i know some surfers have Javascript disabled but 30% - no way is that the whole story.

amazed people turn off Javascript when 99% of modern websites utilize JS for many things.

GA is inaccurate and it slows down your site.

Quote:

Originally Posted by Mutt (Post 19111780)

i know some surfers have Javascript disabled but 30% - no way is that the whole story.

amazed people turn off Javascript when 99% of modern websites utilize JS for many things.

Might be bots hitting your site, might be people hotlinking images (which GA doesn't track since there is no js on the image itself), etc.

A 3rd party hosted solution that relies on client side data can never be as accurate as your own server's stats.

And the problems with GA go even further than that. It's not just users without js they can't track, it's even a significant portion of the users with js they can't track.

Almost all personal firewalls these days have a privacy option/setting/module. Most of those will block tracking sites like GA or at least have an option to do so. That means that those users won't be tracked by GA even if they have js enabled in their browser.

Some users will also use their hosts file to block sites they deem harmful or a privacy risk. Mosts of those hosts files you can find online also included the GA domains. That's even more users GA can't track even if they have js enabled.

The bigger problems occur when people start using onclick GA code on their links. A few months back I noticed the links on Fucking Awesome's sites didn't work for me (js enabled). When I disabled js, they links worked. Their GA code turned out to be the problem:

https://gfy.com/showpost.php?p=18974368&postcount=16

Quote:

Originally Posted by u-Bob (Post 18974368)

You've got an onclick that ends in "return false" on your links so of course your links don't work.

I assume that whatever GA code you've got on there modifies those links (adds tracking stuff or whatever) and then takes care of the redirecting.

Here's the thing: we're nullrouting the GA domains, so the GA code never gets downloaded or executed.

In other words: users that are blocking GA (those with GA domains in the hosts file, a privacy program, a personal firewall with a privacy module/option,...) can't click your links and you probably don't even know you had those visitors if you're relying on GA for stats.

I just did a test on one of their sites and it looks like they are still using that onclick GA code. Their sites, their decision, but personally I don't think it's a good idea to sacrifice what could potentially be a significant portion of your traffic (and income) for some fancy GA traffic stats.

i have a massive difference 30%/40% i checked (JavaScript enabled) and that was less than 10%

Quote:

Originally Posted by Mutt (Post 19111765)

there's a big discrepancy between GA's stats and my own server stats - GA shows 30% less visitors than server stats.

normal?

netstat niggard

Some discrepancy between GA and Server stats is normal and to be expected.

Any tracking that relies on remote transmission of data is inherently less accurate than on-server tracking.

A number of things can influence the discrepancy:

1. Some users may have js disabled, as someone noted in this thread.
2. Server loads a partial page, counts a hit, but the GA snippet is not loaded.
3. GA snipped is loaded, but fails to contact GA server for any given arcane network related reason.
4. etc.

Said this, 30% discrepancy is high, actually higher than anything I have observed myself. <15% would me more in the norm.

I would check a couple of things if I had this % discrepancy, such as:

a. is the snippet a legacy (less reliable) version - there has been a couple of different versions over the years.

b. is the snippet placed very down in the page, i.e. a lot of other things need to load before snipped loads (in theory, the most recent versions of the snippet load asynchronously, which means that their position in the page should not have an impact, but I doubt this always works. *Geek alert - a long discussion could be started here on asynchronous transmission, performance impact of GA in various positions etc.)

c. depending on what server stats you are looking at, there may be 'hits' counted when a mere jpg or some other non-page file is loading, while GA counts a hit per actual page). Make sure you are comparing apples with apples.

d. there may be some pages in your site that do not contain the snippet. For example, if you integrated the snippet in a CMS, and then created some manual landing pages or the like, check if such landing pages have the snippet too.

e. check redirects. if you are using any server side 301 or 302 redirects, EVEN IF THE SNIPPET IS THERE, GA will not read it, because the redirect is performed at server side. The server will in most cases count a hit, GA will not.

good luck!

Import your server logs in piwik. You will get the bells and whistles and it will be more accurate than Google Analytics.