Google Analytics question

On Nov 17, 2007, at 7:03 AM, Kirk McElhearn wrote:

> Does anyone grok Google Analytics? I've inserted the code on my site,
> and the numbers I'm getting from Google are several orders of
> magnitude lower than what my ISP's logs show.

Hmmm. Sounds like a good TidBits article....

I've played with Analytics but not seriously -- my guess is that it's
filtering out a bazillion pageviews from various spiders and other
"unimportant" traffic, using an internal database of such things as a
filter. If I take my analog and awstat results and control for the
hits I get from Google, MSN, et al., it definitely gives me similar
10-1 shifts in headcount.

Best,
Jeff

On 18/11/2007 1:03 AM, "Kirk McElhearn" <kirkmcelhearn.com> spake thus:

> Is Google Analytics flawed? Do I not understand it?

GA will only count the pages that have the tracking code embedded in them,
but even so, an order of magnitude difference in figures seems a bit
unusual. Are there pages showing up in the awstats top downloads list that
aren't showing up in the GA top downloads list?

--
Nigel Stanger, Dunedin, NEW ZEALAND.
http://xri.net/=nigel.stanger

On Nov 17, 2007, at 7:03 AM, Kirk McElhearn wrote:

> Does anyone grok Google Analytics? I've inserted the code on my site,
> and the numbers I'm getting from Google are several orders of
> magnitude lower than what my ISP's logs show. For example, in October,
> I had 13K page views according to Google, but 143K according to my ISP
> (via awstats). Google doesn't count my RSS feed, but that's not the
> difference. (FYI, for another site I run, the same differences show
> up, and the RSS feed was about 10K out of 30K page views; Google only
> shows around 2900 page views.)
>
> Is Google Analytics flawed? Do I not understand it?
>

Google Analytics is by far and away the best tool you can find for
the money (free), and IMHO, it's far and away better than many of the
very expensive paid services. Google Analytics will also give you a
gazillion features that Awstats doesn't have.

Awstats was designed for system administrators, and I've always
suspected the way they count visitors has a lot to do with visitors
being considered relative to bandwidth and server loads. Google
Analytics (which is based on an program called Urchin) is geared
towards advertising and content analysis - the methodology was
refined with this in mind.

Awstats counts all page requests, and considers cgi calls, etc., a
unique request (It does, however, count images and CSS as part of a
page). I'm fairly sure it counts Flash files & mp3s as another
request, but I could be wrong. And Awstats counts any click on a link
to a page a request, whether or not the page fully loads or not -
it's still a drain on the server.

I forget how Awstats counts unique visitors; but in Analytics, you
can set the time frame (I might be wrong, but I think the default is
1 month). Analytics also gives you other very helpful counts -
absolute visitors (people who visited your site for the absolutely
first time), returning visitor, page views per visitor, visits per
visitor, etc. And the conversion tracking features are outstanding.

Something that might account for the discrepency is the "Last Update"
- check the time. Also check the periods that the summaries cover are
the same.

Google Analytics also has different dashboards - I found the
marketing dashboard the most useful - the keywords and conversion
summaries are outrageous; others will want the webmaster or the
executive views.

HTH,

Marilyn

Kirk,

On 2007-11-18, at 01:03, Kirk McElhearn wrote:

> Does anyone grok Google Analytics? I've inserted the code on my site,
> and the numbers I'm getting from Google are several orders of
> magnitude lower than what my ISP's logs show.

It's been a while since I looked at Google Analytics, but isn't the
code you're inserting JavaScript? So any kind of robot or browser
sans JavaScript wouldn't be counted.

When I investigated Analytics I disliked the idea of adding code to
every page on my site which would then contact Google when browsed--
too much of a hack and too much overhead--I stuck with the web server
access log.

On Nov 19, 2007, at 6:43 PM, Jeff Porten wrote:

>> Does anyone grok Google Analytics? I've inserted the code on my site,
>> and the numbers I'm getting from Google are several orders of
>> magnitude lower than what my ISP's logs show.
>
> Hmmm. Sounds like a good TidBits article....
>
> I've played with Analytics but not seriously -- my guess is that it's
> filtering out a bazillion pageviews from various spiders and other
> "unimportant" traffic, using an internal database of such things as a
> filter. If I take my analog and awstat results and control for the
> hits I get from Google, MSN, et al., it definitely gives me similar
> 10-1 shifts in headcount.

My ISP's logs filter out spiders as well...


Kirk

On Nov 19, 2007, at 6:43 PM, Nigel Stanger wrote:

>> Is Google Analytics flawed? Do I not understand it?
>
> GA will only count the pages that have the tracking code embedded in
> them,
> but even so, an order of magnitude difference in figures seems a bit
> unusual. Are there pages showing up in the awstats top downloads
> list that
> aren't showing up in the GA top downloads list?

The code's in the footer of my page, so every page my CMS generates
has it. Yes, the top downloads don't show up in GA anywhere near the
same place. It's all over the place.


Kirk

On 11/19/2007 11:48 AM, "Marilyn Matty" <mmattynyc.rr.com> wrote:

>
> Google Analytics also has different dashboards - I found the
> marketing dashboard the most useful - the keywords and conversion
> summaries are outrageous; others will want the webmaster or the
> executive views.

The problem I have is that while my ISP tells me useful things like what
application was using what user agent, so I can see who's using a news
reader, etc., Google analytics doesn't.

--
John C. Welch

On 20/11/2007 6:43 AM, "Jeff Porten" <civitanjeffporten.com> spake thus:

> I've played with Analytics but not seriously -- my guess is that it's
> filtering out a bazillion pageviews from various spiders and other
> "unimportant" traffic, using an internal database of such things as a
> filter. If I take my analog and awstat results and control for the
> hits I get from Google, MSN, et al., it definitely gives me similar
> 10-1 shifts in headcount.

Interesting, I thought awstats did that kind of filtering already. I know
the code's in there for doing so, as I went looking for it a while back. I
suppose the filtering may be turned off by default?

(I implemented something similar for another project using a Big Ugly Lookup
Table and I wanted to see if there was a nicer way of doing it. To my
surprise, awstats does essentially the same thing. Slightly more
window-dressing than what I did, but effectively still just a Big Ugly
Lookup Table.)

--
Nigel Stanger, Dunedin, NEW ZEALAND.
http://xri.net/=nigel.stanger

On Nov 20, 2007, at 12:32 PM, Tom Robinson wrote:

>> Does anyone grok Google Analytics? I've inserted the code on my site,
>> and the numbers I'm getting from Google are several orders of
>> magnitude lower than what my ISP's logs show.
>
> It's been a while since I looked at Google Analytics, but isn't the
> code you're inserting JavaScript? So any kind of robot or browser
> sans JavaScript wouldn't be counted.

According to what I've read, that would eliminate at most 10% or so
for people who don't have it turned on. GA does report search robots,
so that gets picked up.
>
>
> When I investigated Analytics I disliked the idea of adding code to
> every page on my site which would then contact Google when browsed--
> too much of a hack and too much overhead--I stuck with the web server
> access log.

In my case, I just needed to put it in the page footer; that's one
file that gets added to all of my CMS's pages.


Kirk

Kirk McElhearn wrote:

>... the numbers I'm getting from Google are several orders of
>magnitude lower than what my ISP's logs show. For example, in October,
>I had 13K page views according to Google, but 143K according to my ISP

Perhaps your ISP is like my ISP, which supplies very fancy and
complicated stats that can be sliced and diced in many different
ways. But when you look at the raw data on which they are based, you
realize that the one way which you CANNOT slice them is to
distinguish between browser accesses and bot accesses. Because of
that, the stats are utterly useless for discovering what use _people_
make of the Website.

If Google Analytics is filtering out the visits from bots, I say
"Hurray for them!"

Kirk:

You're not the only one to get results like that. I have several we sites that use Google Analytics and the reports from Google is in the order of 1/10 that of what sever logs state.

Part of the problem is this only works with javascript turned on, but that only account for less than 6 percent.

I contacted Google, but they offered no explanation.

It's still a mystery to me.

Cheers,

tedd

On 12/6/07, tedd <teddsperling.com> wrote:
> You're not the only one to get results like that. I have several we sites that use Google Analytics and the reports from Google is in the order of 1/10 that of what sever logs state.
>
> Part of the problem is this only works with javascript turned on, but that only account for less than 6 percent.

Maybe, maybe not. A huge number of the hits in my server logs are from
bots of various sorts, including a huge number of comment-spam
crawlers. Ugh! None of these guys will trigger 'Analytics.

There's also a good number of connections that seem to be prefetching
requests or reloads or something like that. I'm fairly certain that
Google cleans some of that up.

When I use other similar systems, I get comparable results. I just
don't think web logs are that accurate.

--
Nik

nikinik.net | http://inik.net | http://notions.inik.net