insight in analytics data interpretation

in one day BOOM

[Aircut]

we are working on a new site layout, our site is over 15 years old and the traffic is mostly natural and almost unbelievable stable... for years now we have been getting 10K visitors oer day,

on may 30th we went live with a new site layout. google analytics code was correctly installed in all the pages.

the metrics however are showing 40% decrease in visits (?)...which contradict our server logs....

google analytics also shows that

Pageviews increased in 30%
Pages/Visit increased in 30%
Avg. Time on Site increased in 30%
Bounce Rate decreased in 10%
% New Visits very slightly decreased

we are sure that the tracking code is correctly installed and puzzled with the numbers. anyone might have an insight on how to interpretate the map?

thanks!

[cgrantski]

This is typical when you compare server logs to tagged-generated data.

The most complete list of possible reasons I have found is on The WebTrends Outsider blog site. Search for something like "server log vs javascript tags"

[Aircut]

thank you cgrantski

what i am trying to figure out what are the reasons fro the drop in analytics that corresponds with the day we have undated the site, and has probably have nothing to do with actual traffic coming to the site,,,,

as you can see we managed to improve all other parameters.....according to the analytics....

[cgrantski]

Ah I didn't understand that you already had Google Analytics on the old site and you were comparing old and new Google Analytics data. Sorry.

Google Analytics is telling you that there was a 40% decrease in visits, a 30% increase in page views, and a 30% increase in pages per visit? The arithmetic of those numbers doesn't make sense.

All the other numbers except for the decrease in visit make sense. Your site is apparently more interesting now - people are looking at 30% more pages, spending 30% more time on the site, bouncing less.

The only number that doesn't fit is the decrease in visits. With 30% more pages viewed and 30% increase in pages per visit, your visit count has to be the same before and after. The math doesn't work otherwise.

[Aircut]

thats what i am trying to figure out. why analytics is giving certainly wrong visit traffics.

another glitch is that the site overlay desnt work for the site anymore....

is there anything i can do to rectify the isue? i find the help support of analytics to be complicated and you cant reach a support ticket system....

any leads?

[sullen]

do you mean pageviews for the whole site increased by 30%, or pageviews for individual pages?

It does sound to me like you've missed the code from some pages, or possibly that you *are* down in visitors because many are getting 404 errors (and some log readers would count those as visitors, Google wouldn't).

[Aircut]

yes...30% increase in pageviews for the entire site...

we double checked the code and it is not missing from any of the 250K pages of our site..which in most area is template driven....

since the change, the site overlay of google start behaving wired... and giving us a page not found, webmaster tools and our logs are quite clean form 404....

still puzzled

[sullen]

Are you sure about those figures though? If you divide the total page views by the total visitors, do you get the figure analytics gives for the pages per visit? and the same for the old figures?

[epistable]

One possibility is that the new numbers are accurate and the historic ones were inflated. I'm not sure how the site architecture changed (but I think this is key), but lets say that you previously had the code on a landing page like www.example.com and that most of your visitors ended up going to forums.example.com. Now, if you had installed just the generic code with no modifications, then each visit would be counted twice, because the visitor would be cookied for www.example.com and then for forums.example.com.

Now, let's say that your new site architecture has www.example.com and www.example.com/forums . Now, since there are no longer 2 domains involved, the visitor is counted only once, but the length of their visit is longer, which seems to be what you're seeing in your numbers.

[Aircut]

> One possibility is that the new numbers are accurate and the historic ones were inflated.

I dont think so.... we have Urchin on the site as well and it keeps very close to analyics in numbers. no change there...

the only thing i can think of is that the sub domains filter is not working, we reviewed that and it all seems to be correct, really wired.

[epistable]

oh, it's very interesting that there has been no change in Urchin numbers but there has been in Analytics numbers. What kind of tracking are you doing in Urchin? IP+Useragent or the JS tag? What does your subdomain filter look like, and how is it applied? Feel free to PM me with the site and any other info and I'll take a look.

[Aircut]

thanks! i have just PM you.....